Man1 - ffmpeg-filters.1

A filtergraph has a textual representation, which is recognized by the -filter*/-vf*/*-af* and -filter_complex options in ffmpeg and -vf*/-af* in ffplay, and by the avfilter_graph_parse_ptr() function defined in libavfilter/avfilter.h.

A filterchain consists of a sequence of connected filters, each one connected to the previous one in the sequence. A filterchain is represented by a list of ,-separated filter descriptions.

A filtergraph consists of a sequence of filterchains. A sequence of filterchains is represented by a list of ;-separated filterchain descriptions.

A filter is represented by a string of the form: [/in_link_1/]…[/in_link_N/]/filter_name/@/id/=/arguments/[/out_link_1/]…[/out_link_M/]

filter_name is the name of the filter class of which the described filter is an instance of, and has to be the name of one of the filter classes registered in the program optionally followed by “@/id/. The name of the filter class is optionally followed by a string =/arguments/”.

arguments is a string which contains the parameters used to initialize the filter instance. It may have one of two forms:

• A ’:’-separated list of key=value pairs.
• A ’:’-separated list of value. In this case, the keys are assumed to be the option names in the order they are declared. E.g. the fade filter declares three options in this order Ω- type, start_frame and nb_frames. Then the parameter list in:0:30 means that the value in is assigned to the option type, 0 to start_frame and 30 to nb_frames.
• A ’:’-separated list of mixed direct value and long key=value pairs. The direct value must precede the key=value pairs, and follow the same constraints order of the previous point. The following key=value pairs can be set in any preferred order.

If the option value itself is a list of items (e.g. the format filter takes a list of pixel formats), the items in the list are usually separated by |.

The list of arguments can be quoted using the character ’ as initial and ending mark, and the character \ for escaping the characters within the quoted text; otherwise the argument string is considered terminated when the next special character (belonging to the set []=;,) is encountered.

The name and arguments of the filter are optionally preceded and followed by a list of link labels. A link label allows one to name a link and associate it to a filter output or input pad. The preceding labels in_link_1 … in_link_N, are associated to the filter input pads, the following labels out_link_1 … out_link_M, are associated to the output pads.

When two link labels with the same name are found in the filtergraph, a link between the corresponding input and output pad is created.

If an output pad is not labelled, it is linked by default to the first unlabelled input pad of the next filter in the filterchain. For example in the filterchain

nullsrc, split[L1], [L2]overlay, nullsink

the split filter instance has two output pads, and the overlay filter instance two input pads. The first output pad of split is labelled L1, the first input pad of overlay is labelled L2, and the second output pad of split is linked to the second input pad of overlay, which are both unlabelled.

In a filter description, if the input label of the first filter is not specified, in is assumed; if the output label of the last filter is not specified, out is assumed.

In a complete filterchain all the unlabelled filter input and output pads must be connected. A filtergraph is considered valid if all the filter input and output pads of all the filterchains are connected.

Libavfilter will automatically insert scale filters where format conversion is required. It is possible to specify swscale flags for those automatically inserted scalers by prepending sws_flags==/=flags=/;= to the filtergraph description.

Here is a BNF description of the filtergraph syntax:

<NAME> ::= sequence of alphanumeric characters and _ <FILTER_NAME> ::= <NAME>[“@”<NAME>] <LINKLABEL> ::= “[” <NAME> “]” <LINKLABELS> ::= <LINKLABEL> [<LINKLABELS>] <FILTER_ARGUMENTS> ::= sequence of chars (possibly quoted) <FILTER> ::= [<LINKLABELS>] <FILTER_NAME> [“" <FILTER_ARGUMENTS>] [<LINKLABELS>] <FILTERCHAIN> :: <FILTER> [,<FILTERCHAIN>] <FILTERGRAPH> ::= [sws_flags=<flags>;] <FILTERCHAIN> [;<FILTERGRAPH>]

Filtergraph description composition entails several levels of escaping. See the Quoting and escaping section in the ffmpeg-utils (1) manual for more information about the employed escaping procedure.

A first level escaping affects the content of each filter option value, which may contain the special character : used to separate values, or one of the escaping characters \.

A second level escaping affects the whole filter description, which may contain the escaping characters \ or the special characters [],; used by the filtergraph description.

Finally, when you specify a filtergraph on a shell commandline, you need to perform a third level escaping for the shell special characters contained within it.

For example, consider the following string to be embedded in the drawtext filter description text value:

this is a string: may contain one, or more, special characters

This string contains the special escaping character, and the : special character, so it needs to be escaped in this way:

text=this is a \string\\: may contain one, or more, special characters

A second level of escaping is required when embedding the filter description in a filtergraph description, in order to escape all the filtergraph special characters. Thus the example above becomes:

drawtext=text=this is a \\\string\\\\\: may contain one\, or more\, special characters

(note that in addition to the \ escaping special characters, also , needs to be escaped).

Finally an additional level of escaping is needed when writing the filtergraph description in a shell command, which depends on the escaping rules of the adopted shell. For example, assuming that \ is special and needs to be escaped with another \, the previous string will finally result in:

-vf “drawtext=text=this is a \\\\\\string\\\\\\\\\\: may contain one\\, or more\\, special characters”

A compressor is mainly used to reduce the dynamic range of a signal. Especially modern music is mostly compressed at a high ratio to improve the overall loudness. It’s done to get the highest attention of a listener, fatten the sound and bring more power to the track. If a signal is compressed too much it may sound dull or dead afterwards or it may start to pump (which could be a powerful effect but can also destroy a track completely). The right compression is the key to reach a professional sound and is the high art of mixing and mastering. Because of its complex settings it may take a long time to get the right feeling for this kind of effect.

Compression is done by detecting the volume above a chosen level threshold and dividing it by the factor set with ratio. So if you set the threshold to -12dB and your signal reaches -6dB a ratio of 2:1 will result in a signal at -9dB. Because an exact manipulation of the signal would cause distortion of the waveform the reduction can be levelled over the time. This is done by setting Attack and Release. attack determines how long the signal has to rise above the threshold before any reduction will occur and release sets the time the signal has to fall below the threshold to reduce the reduction again. Shorter signals than the chosen attack time will be left untouched. The overall reduction of the signal can be made up afterwards with the makeup setting. So compressing the peaks of a signal about 6dB and raising the makeup to this level results in a signal twice as loud than the source. To gain a softer entry in the compression the knee flattens the hard edge at the threshold in the range of the chosen decibels.

The filter accepts the following options:

level_in

Set input gain. Default is 1. Range is between 0.015625 and 64.

mode

Set mode of compressor operation. Can be upward or downward. Default is downward.

threshold

If a signal of stream rises above this level it will affect the gain reduction. By default it is 0.125. Range is between 0.00097563 and 1.

ratio

Set a ratio by which the signal is reduced. 1:2 means that if the level rose 4dB above the threshold, it will be only 2dB above after the reduction. Default is 2. Range is between 1 and 20.

attack

Amount of milliseconds the signal has to rise above the threshold before gain reduction starts. Default is 20. Range is between 0.01 and 2000.

release

Amount of milliseconds the signal has to fall below the threshold before reduction is decreased again. Default is 250. Range is between 0.01 and 9000.

makeup

Set the amount by how much signal will be amplified after processing. Default is 1. Range is from 1 to 64.

knee

Curve the sharp knee around the threshold to enter gain reduction more softly. Default is 2.82843. Range is between 1 and 8.

link

Choose if the average level between all channels of input stream or the louder(maximum) channel of input stream affects the reduction. Default is average.

detection

Should the exact signal be taken in case of peak or an RMS one in case of rms. Default is rms which is mostly smoother.

(no term)

How much to use compressed signal in output. Default is 1. Range is between 0 and 1.

Commands

This filter supports the all above options as commands.

Simple audio dynamic range compression/expansion filter.

The filter accepts the following options:

contrast

Set contrast. Default is 33. Allowed range is between 0 and 100.

Copy the input audio source unchanged to the output. This is mainly useful for testing purposes.

Apply cross fade from one input audio stream to another input audio stream. The cross fade is applied for specified duration near the end of first stream.

The filter accepts the following options:

nb_samples, ns

Specify the number of samples for which the cross fade effect has to last. At the end of the cross fade effect the first input audio will be completely silent. Default is 44100.

duration, d

Specify the duration of the cross fade effect. See the Time duration section in the ffmpeg-utils (1) manual for the accepted syntax. By default the duration is determined by nb_samples. If set this option is used instead of nb_samples.

overlap, o

Should first stream end overlap with second stream start. Default is enabled.

curve1

Set curve for cross fade transition for first stream.

curve2

Set curve for cross fade transition for second stream. For description of available curve types see afade filter description.

Examples

• Cross fade from one input to another: ffmpeg -i first.flac -i second.flac -filter_complex acrossfade=d=10:c1=exp:c2=exp output.flac
• Cross fade from one input to another but without overlapping: ffmpeg -i first.flac -i second.flac -filter_complex acrossfade=d=10:o=0:c1=exp:c2=exp output.flac

Split audio stream into several bands.

This filter splits audio stream into two or more frequency ranges. Summing all streams back will give flat output.

The filter accepts the following options:

split

Set split frequencies. Those must be positive and increasing.

order

Set filter order for each band split. This controls filter roll-off or steepness of filter transfer function. Available values are:

2nd

12 dB per octave.

4th

24 dB per octave.

6th

36 dB per octave.

8th

48 dB per octave.

10th

60 dB per octave.

12th

72 dB per octave.

14th

84 dB per octave.

16th

96 dB per octave.

18th

108 dB per octave.

20th

120 dB per octave.

Default is 4th.

level

Set input gain level. Allowed range is from 0 to 1. Default value is 1.

gains

Set output gain for each band. Default value is 1 for all bands.

Examples

• Split input audio stream into two bands (low and high) with split frequency of 1500 Hz, each band will be in separate stream: ffmpeg -i in.flac -filter_complex acrossover=split=1500[LOW][HIGH] -map [LOW] low.wav -map [HIGH] high.wav
• Same as above, but with higher filter order: ffmpeg -i in.flac -filter_complex acrossover=split=1500:order=8th[LOW][HIGH] -map [LOW] low.wav -map [HIGH] high.wav
• Same as above, but also with additional middle band (frequencies between 1500 and 8000): ffmpeg -i in.flac -filter_complex acrossover=split=1500 8000:order=8th[LOW][MID][HIGH] -map [LOW] low.wav -map [MID] mid.wav -map [HIGH] high.wav

Reduce audio bit resolution.

This filter is bit crusher with enhanced functionality. A bit crusher is used to audibly reduce number of bits an audio signal is sampled with. This doesn’t change the bit depth at all, it just produces the effect. Material reduced in bit depth sounds more harsh and digital. This filter is able to even round to continuous values instead of discrete bit depths. Additionally it has a D/C offset which results in different crushing of the lower and the upper half of the signal. An Anti-Aliasing setting is able to produce softer crushing sounds.

Another feature of this filter is the logarithmic mode. This setting switches from linear distances between bits to logarithmic ones. The result is a much more natural sounding crusher which doesn’t gate low signals for example. The human ear has a logarithmic perception, so this kind of crushing is much more pleasant. Logarithmic crushing is also able to get anti-aliased.

The filter accepts the following options:

level_in

Set level in.

level_out

Set level out.

bits

Set bit reduction.

(no term)

Set mixing amount.

mode

Can be linear: lin or logarithmic: log.

(no term)

Set DC.

aa

Set anti-aliasing.

samples

Set sample reduction.

lfo

Enable LFO. By default disabled.

lforange

Set LFO range.

lforate

Set LFO rate.

Commands

This filter supports the all above options as commands.

Delay audio filtering until a given wallclock timestamp. See the cue filter.

Remove impulsive noise from input audio.

Samples detected as impulsive noise are replaced by interpolated samples using autoregressive modelling.

window, w

Set window size, in milliseconds. Allowed range is from 10 to 100. Default value is 55 milliseconds. This sets size of window which will be processed at once.

overlap, o

Set window overlap, in percentage of window size. Allowed range is from 50 to 95. Default value is 75 percent. Setting this to a very high value increases impulsive noise removal but makes whole process much slower.

arorder, a

Set autoregression order, in percentage of window size. Allowed range is from 0 to 25. Default value is 2 percent. This option also controls quality of interpolated samples using neighbour good samples.

threshold, t

Set threshold value. Allowed range is from 1 to 100. Default value is 2. This controls the strength of impulsive noise which is going to be removed. The lower value, the more samples will be detected as impulsive noise.

burst, b

Set burst fusion, in percentage of window size. Allowed range is 0 to 10. Default value is 2. If any two samples detected as noise are spaced less than this value then any sample between those two samples will be also detected as noise.

method, m

Set overlap method. It accepts the following values:

add, a

Select overlap-add method. Even not interpolated samples are slightly changed with this method.

save, s

Select overlap-save method. Not interpolated samples remain unchanged.

Default value is a.

Remove clipped samples from input audio.

Samples detected as clipped are replaced by interpolated samples using autoregressive modelling.

window, w

Set window size, in milliseconds. Allowed range is from 10 to 100. Default value is 55 milliseconds. This sets size of window which will be processed at once.

overlap, o

Set window overlap, in percentage of window size. Allowed range is from 50 to 95. Default value is 75 percent.

arorder, a

Set autoregression order, in percentage of window size. Allowed range is from 0 to 25. Default value is 8 percent. This option also controls quality of interpolated samples using neighbour good samples.

threshold, t

Set threshold value. Allowed range is from 1 to 100. Default value is 10. Higher values make clip detection less aggressive.

hsize, n

Set size of histogram used to detect clips. Allowed range is from 100 to 9999. Default value is 1000. Higher values make clip detection less aggressive.

method, m

Set overlap method. It accepts the following values:

add, a

Select overlap-add method. Even not interpolated samples are slightly changed with this method.

save, s

Select overlap-save method. Not interpolated samples remain unchanged.

Default value is a.

Delay one or more audio channels.

Samples in delayed channel are filled with silence.

The filter accepts the following option:

delays

Set list of delays in milliseconds for each channel separated by ’|’. Unused delays will be silently ignored. If number of given delays is smaller than number of channels all remaining channels will not be delayed. If you want to delay exact number of samples, append ’S’ to number. If you want instead to delay in seconds, append ’s’ to number.

all

Use last set delay for all remaining channels. By default is disabled. This option if enabled changes how option delays is interpreted.

Examples

• Delay first channel by 1.5 seconds, the third channel by 0.5 seconds and leave the second channel (and any other channels that may be present) unchanged. adelay=1500|0|500
• Delay second channel by 500 samples, the third channel by 700 samples and leave the first channel (and any other channels that may be present) unchanged. adelay=0|500S|700S
• Delay all channels by same number of samples: adelay=delays=64S:all=1

Remedy denormals in audio by adding extremely low-level noise.

This filter shall be placed before any filter that can produce denormals.

A description of the accepted parameters follows.

level

Set level of added noise in dB. Default is -351. Allowed range is from -451 to -90.

type

Set type of added noise.

1. Add DC signal.
2. Add AC signal.
3. Add square signal.
4. Add pulse signal.

Default is dc.

Commands

This filter supports the all above options as commands.

Compute derivative/integral of audio stream.

Applying both filters one after another produces original audio.

Apply echoing to the input audio.

Echoes are reflected sound and can occur naturally amongst mountains (and sometimes large buildings) when talking or shouting; digital echo effects emulate this behaviour and are often used to help fill out the sound of a single instrument or vocal. The time difference between the original signal and the reflection is the delay, and the loudness of the reflected signal is the decay. Multiple echoes can have different delays and decays.

A description of the accepted parameters follows.

in_gain

Set input gain of reflected signal. Default is 0.6.

out_gain

Set output gain of reflected signal. Default is 0.3.

delays

Set list of time intervals in milliseconds between original signal and reflections separated by ’|’. Allowed range for each delay is (0 - 90000.0]. Default is 1000.

decays

Set list of loudness of reflected signals separated by ’|’. Allowed range for each decay is (0 - 1.0]. Default is 0.5.

Examples

• Make it sound as if there are twice as many instruments as are actually playing: aecho=0.8:0.88:60:0.4
• If delay is very short, then it sounds like a (metallic) robot playing music: aecho=0.8:0.88:6:0.4
• A longer delay will sound like an open air concert in the mountains: aecho=0.8:0.9:1000:0.3
• Same as above but with one more mountain: aecho=0.8:0.9:1000|1800:0.3|0.25

Audio emphasis filter creates or restores material directly taken from LPs or emphased CDs with different filter curves. E.g. to store music on vinyl the signal has to be altered by a filter first to even out the disadvantages of this recording medium. Once the material is played back the inverse filter has to be applied to restore the distortion of the frequency response.

The filter accepts the following options:

level_in

Set input gain.

level_out

Set output gain.

mode

Set filter mode. For restoring material use reproduction mode, otherwise use production mode. Default is reproduction mode.

type

Set filter type. Selects medium. Can be one of the following:

col

select Columbia.

emi

select EMI.

bsi

select BSI (78RPM).

riaa

select RIAA.

(no term)

select Compact Disc (CD).

50fm

select 50Xs (FM).

75fm

select 75Xs (FM).

50kf

select 50Xs (FM-KF).

75kf

select 75Xs (FM-KF).

Commands

This filter supports the all above options as commands.

Modify an audio signal according to the specified expressions.

This filter accepts one or more expressions (one for each channel), which are evaluated and used to modify a corresponding audio signal.

It accepts the following parameters:

exprs

Set the ’|’-separated expressions list for each separate channel. If the number of input channels is greater than the number of expressions, the last specified expression is used for the remaining output channels.

channel_layout, c

Set output channel layout. If not specified, the channel layout is specified by the number of expressions. If set to same, it will use by default the same input channel layout.

Each expression in exprs can contain the following constants and functions:

ch

channel number of the current expression

(no term)

number of the evaluated sample, starting from 0

(no term)

sample rate

(no term)

time of the evaluated sample expressed in seconds

nb_in_channels

nb_out_channels

input and output number of channels

val(CH)

the value of input channel with number CH

Note: this filter is slow. For faster processing you should use a dedicated filter.

Examples

• Half volume: aeval=val(ch)/2:c=same
• Invert phase of the second channel: aeval=val(0)|-val(1)

An exciter is used to produce high sound that is not present in the original signal. This is done by creating harmonic distortions of the signal which are restricted in range and added to the original signal. An Exciter raises the upper end of an audio signal without simply raising the higher frequencies like an equalizer would do to create a more crisp or brilliant sound.

The filter accepts the following options:

level_in

Set input level prior processing of signal. Allowed range is from 0 to 64. Default value is 1.

level_out

Set output level after processing of signal. Allowed range is from 0 to 64. Default value is 1.

amount

Set the amount of harmonics added to original signal. Allowed range is from 0 to 64. Default value is 1.

drive

Set the amount of newly created harmonics. Allowed range is from 0.1 to 10. Default value is 8.5.

blend

Set the octave of newly created harmonics. Allowed range is from -10 to 10. Default value is 0.

freq

Set the lower frequency limit of producing harmonics in Hz. Allowed range is from 2000 to 12000 Hz. Default is 7500 Hz.

ceil

Set the upper frequency limit of producing harmonics. Allowed range is from 9999 to 20000 Hz. If value is lower than 10000 Hz no limit is applied.

listen

Mute the original signal and output only added harmonics. By default is disabled.

Commands

This filter supports the all above options as commands.

Apply fade-in/out effect to input audio.

A description of the accepted parameters follows.

type, t

Specify the effect type, can be either in for fade-in, or out for a fade-out effect. Default is in.

start_sample, ss

Specify the number of the start sample for starting to apply the fade effect. Default is 0.

nb_samples, ns

Specify the number of samples for which the fade effect has to last. At the end of the fade-in effect the output audio will have the same volume as the input audio, at the end of the fade-out transition the output audio will be silence. Default is 44100.

start_time, st

Specify the start time of the fade effect. Default is 0. The value must be specified as a time duration; see the Time duration section in the ffmpeg-utils (1) manual for the accepted syntax. If set this option is used instead of start_sample.

duration, d

Specify the duration of the fade effect. See the Time duration section in the ffmpeg-utils (1) manual for the accepted syntax. At the end of the fade-in effect the output audio will have the same volume as the input audio, at the end of the fade-out transition the output audio will be silence. By default the duration is determined by nb_samples. If set this option is used instead of nb_samples.

curve

Set curve for fade transition. It accepts the following values:

tri

select triangular, linear slope (default)

qsin

select quarter of sine wave

hsin

select half of sine wave

esin

select exponential sine wave

log

select logarithmic

ipar

select inverted parabola

qua

select quadratic

cub

select cubic

squ

select square root

cbr

select cubic root

par

select parabola

exp

select exponential

iqsin

select inverted quarter of sine wave

ihsin

select inverted half of sine wave

dese

select double-exponential seat

desi

select double-exponential sigmoid

losi

select logistic sigmoid

sinc

select sine cardinal function

isinc

select inverted sine cardinal function

nofade

no fade applied

Commands

This filter supports the all above options as commands.

Examples

• Fade in first 15 seconds of audio: afade=t=in:ss=0:d=15
• Fade out last 25 seconds of a 900 seconds audio: afade=t=out:st=875:d=25

Denoise audio samples with FFT.

A description of the accepted parameters follows.

nr

Set the noise reduction in dB, allowed range is 0.01 to 97. Default value is 12 dB.

nf

Set the noise floor in dB, allowed range is -80 to -20. Default value is -50 dB.

nt

Set the noise type. It accepts the following values:

1. Select white noise.
2. Select vinyl noise.
3. Select shellac noise.
4. Select custom noise, defined in bn option. Default value is white noise.

bn

Set custom band noise for every one of 15 bands. Bands are separated by ’ ’ or ’|’.

rf

Set the residual floor in dB, allowed range is -80 to -20. Default value is -38 dB.

tn

Enable noise tracking. By default is disabled. With this enabled, noise floor is automatically adjusted.

tr

Enable residual tracking. By default is disabled.

om

Set the output mode. It accepts the following values:

1. Pass input unchanged.
2. Pass noise filtered out.
3. Pass only noise. Default value is o.

Commands

This filter supports the following commands:

sample_noise, sn

Start or stop measuring noise profile. Syntax for the command is : start or stop string. After measuring noise profile is stopped it will be automatically applied in filtering.

noise_reduction, nr

Change noise reduction. Argument is single float number. Syntax for the command is : “noise_reduction”

noise_floor, nf

Change noise floor. Argument is single float number. Syntax for the command is : “noise_floor”

output_mode, om

Change output mode operation. Syntax for the command is : i, o or n string.

Apply arbitrary expressions to samples in frequency domain.

real

Set frequency domain real expression for each separate channel separated by ’|’. Default is re. If the number of input channels is greater than the number of expressions, the last specified expression is used for the remaining output channels.

imag

Set frequency domain imaginary expression for each separate channel separated by ’|’. Default is im. Each expression in real and imag can contain the following constants and functions:

sr

sample rate

(no term)

current frequency bin number

nb

number of available bins

ch

channel number of the current expression

chs

number of channels

pts

current frame pts

re

current real part of frequency bin of current channel

im

current imaginary part of frequency bin of current channel

real(b, ch)

Return the value of real part of frequency bin at location (bin,/channel/)

imag(b, ch)

Return the value of imaginary part of frequency bin at location (bin,/channel/)

win_size

Set window size. Allowed range is from 16 to 131072. Default is 4096

win_func

Set window function. Default is hann.

overlap

Set window overlap. If set to 1, the recommended overlap for selected window function will be picked. Default is 0.75.

Examples

• Leave almost only low frequencies in audio: afftfilt=“real=re * (1-clip((b/nb)*b,0,1)):imag=im * (1-clip((b/nb)*b,0,1))”
• Apply robotize effect: afftfilt=“real=hypot(re,im)*sin(0):imag=hypot(re,im)*cos(0):win_size=512:overlap=0.75”
• Apply whisper effect: afftfilt=“real=hypot(re,im)*cos((random(0)*2-1)*2*3.14):imag=hypot(re,im)*sin((random(1)*2-1)*2*3.14):win_size=128:overlap=0.8”

Apply an arbitrary Finite Impulse Response filter.

This filter is designed for applying long FIR filters, up to 60 seconds long.

It can be used as component for digital crossover filters, room equalization, cross talk cancellation, wavefield synthesis, auralization, ambiophonics, ambisonics and spatialization.

This filter uses the streams higher than first one as FIR coefficients. If the non-first stream holds a single channel, it will be used for all input channels in the first stream, otherwise the number of channels in the non-first stream must be same as the number of channels in the first stream.

It accepts the following parameters:

dry

Set dry gain. This sets input gain.

wet

Set wet gain. This sets final output gain.

length

Set Impulse Response filter length. Default is 1, which means whole IR is processed.

gtype

Enable applying gain measured from power of IR. Set which approach to use for auto gain measurement.

none

Do not apply any gain.

peak

select peak gain, very conservative approach. This is default value.

(no term)

select DC gain, limited application.

gn

select gain to noise approach, this is most popular one.

irgain

Set gain to be applied to IR coefficients before filtering. Allowed range is 0 to 1. This gain is applied after any gain applied with gtype option.

irfmt

Set format of IR stream. Can be mono or input. Default is input.

maxir

Set max allowed Impulse Response filter duration in seconds. Default is 30 seconds. Allowed range is 0.1 to 60 seconds.

response

Show IR frequency response, magnitude(magenta), phase(green) and group delay(yellow) in additional video stream. By default it is disabled.

channel

Set for which IR channel to display frequency response. By default is first channel displayed. This option is used only when response is enabled.

size

Set video stream size. This option is used only when response is enabled.

rate

Set video stream frame rate. This option is used only when response is enabled.

minp

Set minimal partition size used for convolution. Default is 8192. Allowed range is from 1 to 32768. Lower values decreases latency at cost of higher CPU usage.

maxp

Set maximal partition size used for convolution. Default is 8192. Allowed range is from 8 to 32768. Lower values may increase CPU usage.

nbirs

Set number of input impulse responses streams which will be switchable at runtime. Allowed range is from 1 to 32. Default is 1.

ir

Set IR stream which will be used for convolution, starting from 0, should always be lower than supplied value by nbirs option. Default is 0. This option can be changed at runtime via commands.

Examples

• Apply reverb to stream using mono IR file as second input, complete command using ffmpeg: ffmpeg -i input.wav -i middle_tunnel_1way_mono.wav -lavfi afir output.wav

Set output format constraints for the input audio. The framework will negotiate the most appropriate format to minimize conversions.

It accepts the following parameters:

sample_fmts, f

A ’|’-separated list of requested sample formats.

sample_rates, r

A ’|’-separated list of requested sample rates.

channel_layouts, cl

A ’|’-separated list of requested channel layouts. See the Channel Layout section in the ffmpeg-utils (1) manual for the required syntax.

If a parameter is omitted, all values are allowed.

Force the output to either unsigned 8-bit or signed 16-bit stereo

aformat=sample_fmts=u8|s16:channel_layouts=stereo

Apply frequency shift to input audio samples.

The filter accepts the following options:

shift

Specify frequency shift. Allowed range is -INT_MAX to INT_MAX. Default value is 0.0.

level

Set output gain applied to final output. Allowed range is from 0.0 to 1.0. Default value is 1.0.

Commands

This filter supports the all above options as commands.

A gate is mainly used to reduce lower parts of a signal. This kind of signal processing reduces disturbing noise between useful signals.

Gating is done by detecting the volume below a chosen level threshold and dividing it by the factor set with ratio. The bottom of the noise floor is set via range. Because an exact manipulation of the signal would cause distortion of the waveform the reduction can be levelled over time. This is done by setting attack and release.

attack determines how long the signal has to fall below the threshold before any reduction will occur and release sets the time the signal has to rise above the threshold to reduce the reduction again. Shorter signals than the chosen attack time will be left untouched.

level_in

Set input level before filtering. Default is 1. Allowed range is from 0.015625 to 64.

mode

Set the mode of operation. Can be upward or downward. Default is downward. If set to upward mode, higher parts of signal will be amplified, expanding dynamic range in upward direction. Otherwise, in case of downward lower parts of signal will be reduced.

range

Set the level of gain reduction when the signal is below the threshold. Default is 0.06125. Allowed range is from 0 to 1. Setting this to 0 disables reduction and then filter behaves like expander.

threshold

If a signal rises above this level the gain reduction is released. Default is 0.125. Allowed range is from 0 to 1.

ratio

Set a ratio by which the signal is reduced. Default is 2. Allowed range is from 1 to 9000.

attack

Amount of milliseconds the signal has to rise above the threshold before gain reduction stops. Default is 20 milliseconds. Allowed range is from 0.01 to 9000.

release

Amount of milliseconds the signal has to fall below the threshold before the reduction is increased again. Default is 250 milliseconds. Allowed range is from 0.01 to 9000.

makeup

Set amount of amplification of signal after processing. Default is 1. Allowed range is from 1 to 64.

knee

Curve the sharp knee around the threshold to enter gain reduction more softly. Default is 2.828427125. Allowed range is from 1 to 8.

detection

Choose if exact signal should be taken for detection or an RMS like one. Default is rms. Can be peak or rms.

link

Choose if the average level between all channels or the louder channel affects the reduction. Default is average. Can be average or maximum.

Commands

This filter supports the all above options as commands.

Apply an arbitrary Infinite Impulse Response filter.

It accepts the following parameters:

zeros, z

Set B/numerator/zeros/reflection coefficients.

poles, p

Set A/denominator/poles/ladder coefficients.

gains, k

Set channels gains.

dry_gain

Set input gain.

wet_gain

Set output gain.

format, f

Set coefficients format.

ll

lattice-ladder function

sf

analog transfer function

tf

digital transfer function

zp

Z-plane zeros/poles, cartesian (default)

pr

Z-plane zeros/poles, polar radians

pd

Z-plane zeros/poles, polar degrees

sp

S-plane zeros/poles

process, r

Set type of processing.

1. direct processing
2. serial processing
3. parallel processing

precision, e

Set filtering precision.

dbl

double-precision floating-point (default)

flt

single-precision floating-point

i32

32-bit integers

i16

16-bit integers

normalize, n

Normalize filter coefficients, by default is enabled. Enabling it will normalize magnitude response at DC to 0dB.

(no term)

How much to use filtered signal in output. Default is 1. Range is between 0 and 1.

response

Show IR frequency response, magnitude(magenta), phase(green) and group delay(yellow) in additional video stream. By default it is disabled.

channel

Set for which IR channel to display frequency response. By default is first channel displayed. This option is used only when response is enabled.

size

Set video stream size. This option is used only when response is enabled.

Coefficients in tf and sf format are separated by spaces and are in ascending order.

Coefficients in zp format are separated by spaces and order of coefficients doesn’t matter. Coefficients in zp format are complex numbers with i imaginary unit.

Different coefficients and gains can be provided for every channel, in such case use ’|’ to separate coefficients or gains. Last provided coefficients will be used for all remaining channels.

Examples

• Apply 2 pole elliptic notch at around 5000Hz for 48000 Hz sample rate: aiir=k=1:z=7.957584807809675810E-1 -2.575128568908332300 3.674839853930788710 -2.57512875289799137 7.957586296317130880E-1:p=1 -2.86950072432325953 3.63022088054647218 -2.28075678147272232 6.361362326477423500E-1:f=tf:r=d
• Same as above but in zp format: aiir=k=0.79575848078096756:z=0.80918701+0.58773007i 0.80918701-0.58773007i 0.80884700+0.58784055i 0.80884700-0.58784055i:p=0.63892345+0.59951235i 0.63892345-0.59951235i 0.79582691+0.44198673i 0.79582691-0.44198673i:f=zp:r=s
• Apply 3-rd order analog normalized Butterworth low-pass filter, using analog transfer function format: aiir=z=1.3057 0 0 0:p=1.3057 2.3892 2.1860 1:f=sf:r=d

The limiter prevents an input signal from rising over a desired threshold. This limiter uses lookahead technology to prevent your signal from distorting. It means that there is a small delay after the signal is processed. Keep in mind that the delay it produces is the attack time you set.

The filter accepts the following options:

level_in

Set input gain. Default is 1.

level_out

Set output gain. Default is 1.

limit

Don’t let signals above this level pass the limiter. Default is 1.

attack

The limiter will reach its attenuation level in this amount of time in milliseconds. Default is 5 milliseconds.

release

Come back from limiting to attenuation 1.0 in this amount of milliseconds. Default is 50 milliseconds.

asc

When gain reduction is always needed ASC takes care of releasing to an average reduction level rather than reaching a reduction of 0 in the release time.

asc_level

Select how much the release time is affected by ASC, 0 means nearly no changes in release time while 1 produces higher release times.

level

Auto level output signal. Default is enabled. This normalizes audio back to 0dB if enabled.

Depending on picked setting it is recommended to upsample input 2x or 4x times with aresample before applying this filter.

Apply a two-pole all-pass filter with central frequency (in Hz) frequency, and filter-width width. An all-pass filter changes the audio’s frequency to phase relationship without changing its frequency to amplitude relationship.

The filter accepts the following options:

frequency, f

Set frequency in Hz.

width_type, t

Set method to specify band-width of filter.

1. Hz
2. Q-Factor
3. octave
4. slope
5. kHz

width, w

Specify the band-width of a filter in width_type units.

mix, m

How much to use filtered signal in output. Default is 1. Range is between 0 and 1.

channels, c

Specify which channels to filter, by default all available are filtered.

normalize, n

Normalize biquad coefficients, by default is disabled. Enabling it will normalize magnitude response at DC to 0dB.

order, o

Set the filter order, can be 1 or 2. Default is 2.

transform, a

Set transform type of IIR filter.

precision, r

Set precison of filtering.

auto

Pick automatic sample format depending on surround filters.

s16

Always use signed 16-bit.

s32

Always use signed 32-bit.

f32

Always use float 32-bit.

f64

Always use float 64-bit.

nil

Commands

This filter supports the following commands:

frequency, f

Change allpass frequency. Syntax for the command is : “frequency”

width_type, t

Change allpass width_type. Syntax for the command is

"/width_type/"

width, w

Change allpass width. Syntax for the command is : “width”

mix, m

Change allpass mix. Syntax for the command is : “mix”

Loop audio samples.

The filter accepts the following options:

loop

Set the number of loops. Setting this value to -1 will result in infinite loops. Default is 0.

size

Set maximal number of samples. Default is 0.

start

Set first sample of loop. Default is 0.

Merge two or more audio streams into a single multi-channel stream.

The filter accepts the following options:

inputs

Set the number of inputs. Default is 2.

If the channel layouts of the inputs are disjoint, and therefore compatible, the channel layout of the output will be set accordingly and the channels will be reordered as necessary. If the channel layouts of the inputs are not disjoint, the output will have all the channels of the first input then all the channels of the second input, in that order, and the channel layout of the output will be the default value corresponding to the total number of channels.

For example, if the first input is in 2.1 (FL+FR+LF) and the second input is FC+BL+BR, then the output will be in 5.1, with the channels in the following order: a1, a2, b1, a3, b2, b3 (a1 is the first channel of the first input, b1 is the first channel of the second input).

On the other hand, if both input are in stereo, the output channels will be in the default order: a1, a2, b1, b2, and the channel layout will be arbitrarily set to 4.0, which may or may not be the expected value.

All inputs must have the same sample rate, and format.

If inputs do not have the same duration, the output will stop with the shortest.

Examples

• Merge two mono files into a stereo stream: amovie=left.wav [l] ; amovie=right.mp3 [r] ; [l] [r] amerge
• Multiple merges assuming 1 video stream and 6 audio streams in input.mkv: ffmpeg -i input.mkv -filter_complex “[0:1][0:2][0:3][0:4][0:5][0:6] amerge=inputs=6” -c:a pcm_s16le output.mkv

Mixes multiple audio inputs into a single output.

Note that this filter only supports float samples (the amerge and pan audio filters support many formats). If the amix input has integer samples then aresample will be automatically inserted to perform the conversion to float samples.

For example

ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex amix=inputs=3:duration=first:dropout_transition=3 OUTPUT

will mix 3 input audio streams to a single output with the same duration as the first input and a dropout transition time of 3 seconds.

It accepts the following parameters:

inputs

The number of inputs. If unspecified, it defaults to 2.

duration

How to determine the end-of-stream.

longest

The duration of the longest input. (default)

shortest

The duration of the shortest input.

first

The duration of the first input.

dropout_transition

The transition time, in seconds, for volume renormalization when an input stream ends. The default value is 2 seconds.

weights

Specify weight of each input audio stream as sequence. Each weight is separated by space. By default all inputs have same weight.

normalize

Always scale inputs instead of only doing summation of samples. Beware of heavy clipping if inputs are not normalized prior or after filtering by this filter if this option is disabled. By default is enabled.

Commands

This filter supports the following commands:

weights

sum

Syntax is same as option with same name.

Multiply first audio stream with second audio stream and store result in output audio stream. Multiplication is done by multiplying each sample from first stream with sample at same position from second stream.

With this element-wise multiplication one can create amplitude fades and amplitude modulations.

High-order parametric multiband equalizer for each channel.

It accepts the following parameters:

params

This option string is in format: “c/chn/ f=/cf/ w=/w/ g=/g/ t=/f/ | …” Each equalizer band is separated by ’|’.

chn

Set channel number to which equalization will be applied. If input doesn’t have that channel the entry is ignored.

(no term)

Set central frequency for band. If input doesn’t have that frequency the entry is ignored.

(no term)

Set band width in Hertz.

(no term)

Set band gain in dB.

(no term)

Set filter type for band, optional, can be:

1. Butterworth, this is default.
2. Chebyshev type 1.
3. Chebyshev type 2.

curves

With this option activated frequency response of anequalizer is displayed in video stream.

size

Set video stream size. Only useful if curves option is activated.

mgain

Set max gain that will be displayed. Only useful if curves option is activated. Setting this to a reasonable value makes it possible to display gain which is derived from neighbour bands which are too close to each other and thus produce higher gain when both are activated.

fscale

Set frequency scale used to draw frequency response in video output. Can be linear or logarithmic. Default is logarithmic.

colors

Set color for each channel curve which is going to be displayed in video stream. This is list of color names separated by space or by ’|’. Unrecognised or missing colors will be replaced by white color.

Examples

• Lower gain by 10 of central frequency 200Hz and width 100 Hz for first 2 channels using Chebyshev type 1 filter: anequalizer=c0 f=200 w=100 g=-10 t=1|c1 f=200 w=100 g=-10 t=1

Commands

This filter supports the following commands:

change

Alter existing filter parameters. Syntax for the commands is

"/fN/|f=/freq/|w=/width/|g=/gain/" /fN/ is existing filter number,

starting from 0, if no such filter is available error is returned. freq set new frequency parameter. width set new width parameter in Hertz. gain set new gain parameter in dB. Full filter invocation with asendcmd may look like this: asendcmd=c=’4.0 anequalizer change 0|f=200|w=50|g=1’,anequalizer=…

Reduce broadband noise in audio samples using Non-Local Means algorithm.

Each sample is adjusted by looking for other samples with similar contexts. This context similarity is defined by comparing their surrounding patches of size p. Patches are searched in an area of r around the sample.

The filter accepts the following options:

1. Set denoising strength. Allowed range is from 0.00001 to 10. Default value is 0.00001.
2. Set patch radius duration. Allowed range is from 1 to 100 milliseconds. Default value is 2 milliseconds.
3. Set research radius duration. Allowed range is from 2 to 300 milliseconds. Default value is 6 milliseconds.
4. Set the output mode. It accepts the following values:
   1. Pass input unchanged.
   2. Pass noise filtered out.
   3. Pass only noise. Default value is o.
5. Set smooth factor. Default value is 11. Allowed range is from 1 to 15.

Commands

This filter supports the all above options as commands.

Apply Normalized Least-Mean-Squares algorithm to the first audio stream using the second audio stream.

This adaptive filter is used to mimic a desired filter by finding the filter coefficients that relate to producing the least mean square of the error signal (difference between the desired, 2nd input audio stream and the actual signal, the 1st input audio stream).

A description of the accepted options follows.

order

Set filter order.

mu

Set filter mu.

eps

Set the filter eps.

leakage

Set the filter leakage.

out_mode

It accepts the following values:

1. Pass the 1st input.
2. Pass the 2nd input.
3. Pass filtered samples.
4. Pass difference between desired and filtered samples. Default value is o.

Examples

• One of many usages of this filter is noise reduction, input audio is filtered with same samples that are delayed by fixed amount, one such example for stereo audio is: asplit[a][b],[a]adelay=32S|32S[a],[b][a]anlms=order=128:leakage=0.0005:mu=.5:out_mode=o

Commands

This filter supports the same commands as options, excluding option order.

Pass the audio source unchanged to the output.

Pad the end of an audio stream with silence.

This can be used together with ffmpeg -shortest to extend audio streams to the same length as the video stream.

A description of the accepted options follows.

packet_size

Set silence packet size. Default value is 4096.

pad_len

Set the number of samples of silence to add to the end. After the value is reached, the stream is terminated. This option is mutually exclusive with whole_len.

whole_len

Set the minimum total number of samples in the output audio stream. If the value is longer than the input audio length, silence is added to the end, until the value is reached. This option is mutually exclusive with pad_len.

pad_dur

Specify the duration of samples of silence to add. See the Time duration section in the ffmpeg-utils (1) manual for the accepted syntax. Used only if set to non-zero value.

whole_dur

Specify the minimum total duration in the output audio stream. See the Time duration section in the ffmpeg-utils (1) manual for the accepted syntax. Used only if set to non-zero value. If the value is longer than the input audio length, silence is added to the end, until the value is reached. This option is mutually exclusive with pad_dur

If neither the pad_len nor the whole_len nor pad_dur nor whole_dur option is set, the filter will add silence to the end of the input stream indefinitely.

Examples

• Add 1024 samples of silence to the end of the input: apad=pad_len=1024
• Make sure the audio output will contain at least 10000 samples, pad the input with silence if required: apad=whole_len=10000
• Use ffmpeg to pad the audio input with silence, so that the video stream will always result the shortest and will be converted until the end in the output file when using the shortest option: ffmpeg -i VIDEO -i AUDIO -filter_complex “[1:0]apad” -shortest OUTPUT

Add a phasing effect to the input audio.

A phaser filter creates series of peaks and troughs in the frequency spectrum. The position of the peaks and troughs are modulated so that they vary over time, creating a sweeping effect.

A description of the accepted parameters follows.

in_gain

Set input gain. Default is 0.4.

out_gain

Set output gain. Default is 0.74

delay

Set delay in milliseconds. Default is 3.0.

decay

Set decay. Default is 0.4.

speed

Set modulation speed in Hz. Default is 0.5.

type

Set modulation type. Default is triangular. It accepts the following values:

triangular, t

sinusoidal, s

Apply phase shift to input audio samples.

The filter accepts the following options:

shift

Specify phase shift. Allowed range is from -1.0 to 1.0. Default value is 0.0.

level

Set output gain applied to final output. Allowed range is from 0.0 to 1.0. Default value is 1.0.

Commands

This filter supports the all above options as commands.

Audio pulsator is something between an autopanner and a tremolo. But it can produce funny stereo effects as well. Pulsator changes the volume of the left and right channel based on a LFO (low frequency oscillator) with different waveforms and shifted phases. This filter have the ability to define an offset between left and right channel. An offset of 0 means that both LFO shapes match each other. The left and right channel are altered equally - a conventional tremolo. An offset of 50% means that the shape of the right channel is exactly shifted in phase (or moved backwards about half of the frequency) - pulsator acts as an autopanner. At 1 both curves match again. Every setting in between moves the phase shift gapless between all stages and produces some bypassing sounds with sine and triangle waveforms. The more you set the offset near 1 (starting from the 0.5) the faster the signal passes from the left to the right speaker.

The filter accepts the following options:

level_in

Set input gain. By default it is 1. Range is [0.015625 - 64].

level_out

Set output gain. By default it is 1. Range is [0.015625 - 64].

mode

Set waveform shape the LFO will use. Can be one of: sine, triangle, square, sawup or sawdown. Default is sine.

amount

Set modulation. Define how much of original signal is affected by the LFO.

offset_l

Set left channel offset. Default is 0. Allowed range is [0 - 1].

offset_r

Set right channel offset. Default is 0.5. Allowed range is [0 - 1].

width

Set pulse width. Default is 1. Allowed range is [0 - 2].

timing

Set possible timing mode. Can be one of: bpm, ms or hz. Default is hz.

bpm

Set bpm. Default is 120. Allowed range is [30 - 300]. Only used if timing is set to bpm.

ms

Set ms. Default is 500. Allowed range is [10 - 2000]. Only used if timing is set to ms.

hz

Set frequency in Hz. Default is 2. Allowed range is [0.01 - 100]. Only used if timing is set to hz.

Resample the input audio to the specified parameters, using the libswresample library. If none are specified then the filter will automatically convert between its input and output.

This filter is also able to stretch/squeeze the audio data to make it match the timestamps or to inject silence / cut out audio to make it match the timestamps, do a combination of both or do neither.

The filter accepts the syntax [/sample_rate/:]/resampler_options/, where sample_rate expresses a sample rate and resampler_options is a list of key/=/value pairs, separated by :. See the Resampler Options section in the ffmpeg-resampler (1) manual for the complete list of supported options.

Examples

• Resample the input audio to 44100Hz: aresample=44100
• Stretch/squeeze samples to the given timestamps, with a maximum of 1000 samples per second compensation: aresample=async=1000

Reverse an audio clip.

Warning: This filter requires memory to buffer the entire clip, so trimming is suggested.

Examples

• Take the first 5 seconds of a clip, and reverse it. atrim=end=5,areverse

Reduce noise from speech using Recurrent Neural Networks.

This filter accepts the following options:

model, m

Set train model file to load. This option is always required.

(no term)

Set how much to mix filtered samples into final output. Allowed range is from -1 to 1. Default value is 1. Negative values are special, they set how much to keep filtered noise in the final filter output. Set this option to -1 to hear actual noise removed from input signal.

Commands

This filter supports the all above options as commands.

Set the number of samples per each output audio frame.

The last output packet may contain a different number of samples, as the filter will flush all the remaining samples when the input audio signals its end.

The filter accepts the following options:

nb_out_samples, n

Set the number of frames per each output audio frame. The number is intended as the number of samples per each channel. Default value is 1024.

pad, p

If set to 1, the filter will pad the last audio frame with zeroes, so that the last frame will contain the same number of samples as the previous ones. Default value is 1.

For example, to set the number of per-frame samples to 1234 and disable padding for the last frame, use:

asetnsamples=n=1234:p=0

Set the sample rate without altering the PCM data. This will result in a change of speed and pitch.

The filter accepts the following options:

sample_rate, r

Set the output sample rate. Default is 44100 Hz.

Show a line containing various information for each input audio frame. The input audio is not modified.

The shown line contains a sequence of key/value pairs of the form key:/value/.

The following values are shown in the output:

1. The (sequential) number of the input frame, starting from 0.
2. The presentation timestamp of the input frame, in time base units; the time base depends on the filter input pad, and is usually 1//sample_rate/.
3. The presentation timestamp of the input frame in seconds.
4. position of the frame in the input stream, -1 if this information in unavailable and/or meaningless (for example in case of synthetic audio)
5. The sample format.
6. The channel layout.
7. The sample rate for the audio frame.
8. The number of samples (per channel) in the frame.
9. The Adler-32 checksum (printed in hexadecimal) of the audio data. For planar audio, the data is treated as if all the planes were concatenated.
10. A list of Adler-32 checksums for each data plane.

Apply audio soft clipping.

Soft clipping is a type of distortion effect where the amplitude of a signal is saturated along a smooth curve, rather than the abrupt shape of hard-clipping.

This filter accepts the following options:

type

Set type of soft-clipping. It accepts the following values:

hard

tanh

atan

cubic

exp

alg

quintic

sin

erf

threshold

Set threshold from where to start clipping. Default value is 0dB or 1.

output

Set gain applied to output. Default value is 0dB or 1.

param

Set additional parameter which controls sigmoid function.

oversample

Set oversampling factor.

Commands

This filter supports the all above options as commands.

Automatic Speech Recognition

This filter uses PocketSphinx for speech recognition. To enable compilation of this filter, you need to configure FFmpeg with --enable-pocketsphinx.

It accepts the following options:

rate

Set sampling rate of input audio. Defaults is 16000. This need to match speech models, otherwise one will get poor results.

hmm

Set dictionary containing acoustic model files.

dict

Set pronunciation dictionary.

lm

Set language model file.

lmctl

Set language model set.

lmname

Set which language model to use.

logfn

Set output for log messages.

The filter exports recognized speech as the frame metadata lavfi.asr.text.

Display time domain statistical information about the audio channels. Statistics are calculated and displayed for each audio channel and, where applicable, an overall figure is also given.

It accepts the following option:

length

Short window length in seconds, used for peak and trough RMS measurement. Default is 0.05 (50 milliseconds). Allowed range is [0.01 - 10].

metadata

Set metadata injection. All the metadata keys are prefixed with lavfi.astats.X, where X is channel number starting from 1 or string Overall. Default is disabled. Available keys for each channel are: DC_offset Min_level Max_level Min_difference Max_difference Mean_difference RMS_difference Peak_level RMS_peak RMS_trough Crest_factor Flat_factor Peak_count Noise_floor Noise_floor_count Bit_depth Dynamic_range Zero_crossings Zero_crossings_rate Number_of_NaNs Number_of_Infs Number_of_denormals and for Overall: DC_offset Min_level Max_level Min_difference Max_difference Mean_difference RMS_difference Peak_level RMS_level RMS_peak RMS_trough Flat_factor Peak_count Noise_floor Noise_floor_count Bit_depth Number_of_samples Number_of_NaNs Number_of_Infs Number_of_denormals For example full key look like this lavfi.astats.1.DC_offset or this lavfi.astats.Overall.Peak_count. For description what each key means read below.

reset

Set number of frame after which stats are going to be recalculated. Default is disabled.

measure_perchannel

Select the entries which need to be measured per channel. The metadata keys can be used as flags, default is all which measures everything. none disables all per channel measurement.

measure_overall

Select the entries which need to be measured overall. The metadata keys can be used as flags, default is all which measures everything. none disables all overall measurement.

A description of each shown parameter follows:

DC offset

Mean amplitude displacement from zero.

Min level

Minimal sample level.

Max level

Maximal sample level.

Min difference

Minimal difference between two consecutive samples.

Max difference

Maximal difference between two consecutive samples.

Mean difference

Mean difference between two consecutive samples. The average of each difference between two consecutive samples.

RMS difference

Root Mean Square difference between two consecutive samples.

Peak level dB

RMS level dB

Standard peak and RMS level measured in dBFS.

RMS peak dB

RMS trough dB

Peak and trough values for RMS level measured over a short window.

Crest factor

Standard ratio of peak to RMS level (note: not in dB).

Flat factor

Flatness (i.e. consecutive samples with the same value) of the signal at its peak levels (i.e. either Min level or Max level).

Peak count

Number of occasions (not the number of samples) that the signal attained either Min level or Max level.

Noise floor dB

Minimum local peak measured in dBFS over a short window.

Noise floor count

Number of occasions (not the number of samples) that the signal attained Noise floor.

Bit depth

Overall bit depth of audio. Number of bits used for each sample.

Dynamic range

Measured dynamic range of audio in dB.

Zero crossings

Number of points where the waveform crosses the zero level axis.

Zero crossings rate

Rate of Zero crossings and number of audio samples.

Boost subwoofer frequencies.

The filter accepts the following options:

dry

Set dry gain, how much of original signal is kept. Allowed range is from 0 to 1. Default value is 0.7.

wet

Set wet gain, how much of filtered signal is kept. Allowed range is from 0 to 1. Default value is 0.7.

decay

Set delay line decay gain value. Allowed range is from 0 to 1. Default value is 0.7.

feedback

Set delay line feedback gain value. Allowed range is from 0 to 1. Default value is 0.9.

cutoff

Set cutoff frequency in Hertz. Allowed range is 50 to 900. Default value is 100.

slope

Set slope amount for cutoff frequency. Allowed range is 0.0001 to 1. Default value is 0.5.

delay

Set delay. Allowed range is from 1 to 100. Default value is 20.

Commands

This filter supports the all above options as commands.

Cut subwoofer frequencies.

This filter allows to set custom, steeper roll off than highpass filter, and thus is able to more attenuate frequency content in stop-band.

The filter accepts the following options:

cutoff

Set cutoff frequency in Hertz. Allowed range is 2 to 200. Default value is 20.

order

Set filter order. Available values are from 3 to 20. Default value is 10.

level

Set input gain level. Allowed range is from 0 to 1. Default value is 1.

Commands

This filter supports the all above options as commands.

Cut super frequencies.

The filter accepts the following options:

cutoff

Set cutoff frequency in Hertz. Allowed range is 20000 to 192000. Default value is 20000.

order

Set filter order. Available values are from 3 to 20. Default value is 10.

level

Set input gain level. Allowed range is from 0 to 1. Default value is 1.

Commands

This filter supports the all above options as commands.

Apply high order Butterworth band-pass filter.

The filter accepts the following options:

centerf

Set center frequency in Hertz. Allowed range is 2 to 999999. Default value is 1000.

order

Set filter order. Available values are from 4 to 20. Default value is 4.

qfactor

Set Q-factor. Allowed range is from 0.01 to 100. Default value is 1.

level

Set input gain level. Allowed range is from 0 to 2. Default value is 1.

Commands

This filter supports the all above options as commands.

Apply high order Butterworth band-stop filter.

The filter accepts the following options:

centerf

Set center frequency in Hertz. Allowed range is 2 to 999999. Default value is 1000.

order

Set filter order. Available values are from 4 to 20. Default value is 4.

qfactor

Set Q-factor. Allowed range is from 0.01 to 100. Default value is 1.

level

Set input gain level. Allowed range is from 0 to 2. Default value is 1.

Commands

This filter supports the all above options as commands.

Adjust audio tempo.

The filter accepts exactly one parameter, the audio tempo. If not specified then the filter will assume nominal 1.0 tempo. Tempo must be in the [0.5, 100.0] range.

Note that tempo greater than 2 will skip some samples rather than blend them in. If for any reason this is a concern it is always possible to daisy-chain several instances of atempo to achieve the desired product tempo.

Examples

• Slow down audio to 80% tempo: atempo=0.8
• To speed up audio to 300% tempo: atempo=3
• To speed up audio to 300% tempo by daisy-chaining two atempo instances: atempo=sqrt(3),atempo=sqrt(3)

Commands

This filter supports the following commands:

tempo

Change filter tempo scale factor. Syntax for the command is : “tempo”

Trim the input so that the output contains one continuous subpart of the input.

It accepts the following parameters:

start

Timestamp (in seconds) of the start of the section to keep. I.e. the audio sample with the timestamp start will be the first sample in the output.

end

Specify time of the first audio sample that will be dropped, i.e. the audio sample immediately preceding the one with the timestamp end will be the last sample in the output.

start_pts

Same as start, except this option sets the start timestamp in samples instead of seconds.

end_pts

Same as end, except this option sets the end timestamp in samples instead of seconds.

duration

The maximum duration of the output in seconds.

start_sample

The number of the first sample that should be output.

end_sample

The number of the first sample that should be dropped.

start, end, and duration are expressed as time duration specifications; see the Time duration section in the ffmpeg-utils (1) manual.

Note that the first two sets of the start/end options and the duration option look at the frame timestamp, while the _sample options simply count the samples that pass through the filter. So start/end_pts and start/end_sample will give different results when the timestamps are wrong, inexact or do not start at zero. Also note that this filter does not modify the timestamps. If you wish to have the output timestamps start at zero, insert the asetpts filter after the atrim filter.

If multiple start or end options are set, this filter tries to be greedy and keep all samples that match at least one of the specified constraints. To keep only the part that matches all the constraints at once, chain multiple atrim filters.

The defaults are such that all the input is kept. So it is possible to set e.g. just the end values to keep everything before the specified time.

Examples:

• Drop everything except the second minute of input: ffmpeg -i INPUT -af atrim=60:120
• Keep only the first 1000 samples: ffmpeg -i INPUT -af atrim=end_sample=1000

Calculate normalized cross-correlation between two input audio streams.

Resulted samples are always between -1 and 1 inclusive. If result is 1 it means two input samples are highly correlated in that selected segment. Result 0 means they are not correlated at all. If result is -1 it means two input samples are out of phase, which means they cancel each other.

The filter accepts the following options:

size

Set size of segment over which cross-correlation is calculated. Default is 256. Allowed range is from 2 to 131072.

algo

Set algorithm for cross-correlation. Can be slow or fast. Default is slow. Fast algorithm assumes mean values over any given segment are always zero and thus need much less calculations to make. This is generally not true, but is valid for typical audio streams.

Examples

• Calculate correlation between channels in stereo audio stream: ffmpeg -i stereo.wav -af channelsplit,axcorrelate=size=1024:algo=fast correlation.wav

Apply a two-pole Butterworth band-pass filter with central frequency frequency, and (3dB-point) band-width width. The csg option selects a constant skirt gain (peak gain = Q) instead of the default: constant 0dB peak gain. The filter roll off at 6dB per octave (20dB per decade).

The filter accepts the following options:

frequency, f

Set the filter’s central frequency. Default is 3000.

csg

Constant skirt gain if set to 1. Defaults to 0.

width_type, t

Set method to specify band-width of filter.

1. Hz
2. Q-Factor
3. octave
4. slope
5. kHz

width, w

Specify the band-width of a filter in width_type units.

mix, m

How much to use filtered signal in output. Default is 1. Range is between 0 and 1.

channels, c

Specify which channels to filter, by default all available are filtered.

normalize, n

Normalize biquad coefficients, by default is disabled. Enabling it will normalize magnitude response at DC to 0dB.

transform, a

Set transform type of IIR filter.

precision, r

Set precison of filtering.

auto

Pick automatic sample format depending on surround filters.

s16

Always use signed 16-bit.

s32

Always use signed 32-bit.

f32

Always use float 32-bit.

f64

Always use float 64-bit.

nil

Commands

This filter supports the following commands:

frequency, f

Change bandpass frequency. Syntax for the command is : “frequency”

width_type, t

Change bandpass width_type. Syntax for the command is

"/width_type/"

width, w

Change bandpass width. Syntax for the command is : “width”

mix, m

Change bandpass mix. Syntax for the command is : “mix”

Apply a two-pole Butterworth band-reject filter with central frequency frequency, and (3dB-point) band-width width. The filter roll off at 6dB per octave (20dB per decade).

The filter accepts the following options:

frequency, f

Set the filter’s central frequency. Default is 3000.

width_type, t

Set method to specify band-width of filter.

1. Hz
2. Q-Factor
3. octave
4. slope
5. kHz

width, w

Specify the band-width of a filter in width_type units.

mix, m

How much to use filtered signal in output. Default is 1. Range is between 0 and 1.

channels, c

Specify which channels to filter, by default all available are filtered.

normalize, n

Normalize biquad coefficients, by default is disabled. Enabling it will normalize magnitude response at DC to 0dB.

transform, a

Set transform type of IIR filter.

precision, r

Set precison of filtering.

auto

Pick automatic sample format depending on surround filters.

s16

Always use signed 16-bit.

s32

Always use signed 32-bit.

f32

Always use float 32-bit.

f64

Always use float 64-bit.

nil

Commands

This filter supports the following commands:

frequency, f

Change bandreject frequency. Syntax for the command is

"/frequency/"

width_type, t

Change bandreject width_type. Syntax for the command is : “width_type”

width, w

Change bandreject width. Syntax for the command is : “width”

mix, m

Change bandreject mix. Syntax for the command is : “mix”

Boost or cut the bass (lower) frequencies of the audio using a two-pole shelving filter with a response similar to that of a standard hi-fi’s tone-controls. This is also known as shelving equalisation (EQ).

The filter accepts the following options:

gain, g

Give the gain at 0 Hz. Its useful range is about -20 (for a large cut) to +20 (for a large boost). Beware of clipping when using a positive gain.

frequency, f

Set the filter’s central frequency and so can be used to extend or reduce the frequency range to be boosted or cut. The default value is 100 Hz.

width_type, t

Set method to specify band-width of filter.

1. Hz
2. Q-Factor
3. octave
4. slope
5. kHz

width, w

Determine how steep is the filter’s shelf transition.

poles, p

Set number of poles. Default is 2.

mix, m

How much to use filtered signal in output. Default is 1. Range is between 0 and 1.

channels, c

Specify which channels to filter, by default all available are filtered.

normalize, n

Normalize biquad coefficients, by default is disabled. Enabling it will normalize magnitude response at DC to 0dB.

transform, a

Set transform type of IIR filter.

precision, r

Set precison of filtering.

auto

Pick automatic sample format depending on surround filters.

s16

Always use signed 16-bit.

s32

Always use signed 32-bit.

f32

Always use float 32-bit.

f64

Always use float 64-bit.

nil

Commands

This filter supports the following commands:

frequency, f

Change bass frequency. Syntax for the command is : “frequency”

width_type, t

Change bass width_type. Syntax for the command is : “width_type”

width, w

Change bass width. Syntax for the command is : “width”

gain, g

Change bass gain. Syntax for the command is : “gain”

mix, m

Change bass mix. Syntax for the command is : “mix”

Apply a biquad IIR filter with the given coefficients. Where b0, b1, b2 and a0, a1, a2 are the numerator and denominator coefficients respectively. and channels, c specify which channels to filter, by default all available are filtered.

Commands

This filter supports the following commands:

a0

a1

a2

b0

b1

b2

Change biquad parameter. Syntax for the command is : “value”

mix, m

How much to use filtered signal in output. Default is 1. Range is between 0 and 1.

channels, c

Specify which channels to filter, by default all available are filtered.

normalize, n

Normalize biquad coefficients, by default is disabled. Enabling it will normalize magnitude response at DC to 0dB.

transform, a

Set transform type of IIR filter.

precision, r

Set precison of filtering.

auto

Pick automatic sample format depending on surround filters.

s16

Always use signed 16-bit.

s32

Always use signed 32-bit.

f32

Always use float 32-bit.

f64

Always use float 64-bit.

nil

Bauer stereo to binaural transformation, which improves headphone listening of stereo audio records.

To enable compilation of this filter you need to configure FFmpeg with --enable-libbs2b.

It accepts the following parameters:

profile

Pre-defined crossfeed level.

default

Default level (fcut=700, feed=50).

cmoy

Chu Moy circuit (fcut=700, feed=60).

jmeier

Jan Meier circuit (fcut=650, feed=95).

fcut

Cut frequency (in Hz).

feed

Feed level (in Hz).

Remap input channels to new locations.

It accepts the following parameters:

map

Map channels from input to output. The argument is a ’|’-separated list of mappings, each in the in_channel=/-/=out_channel or in_channel form. in_channel can be either the name of the input channel (e.g. FL for front left) or its index in the input channel layout. out_channel is the name of the output channel or its index in the output channel layout. If out_channel is not given then it is implicitly an index, starting with zero and increasing by one for each mapping.

channel_layout

The channel layout of the output stream.

If no mapping is present, the filter will implicitly map input channels to output channels, preserving indices.

Examples

• For example, assuming a 5.1+downmix input MOV file, ffmpeg -i in.mov -filter channelmap=map=DL-FL|DR-FR out.wav will create an output WAV file tagged as stereo from the downmix channels of the input.
• To fix a 5.1 WAV improperly encoded in AAC’s native channel order ffmpeg -i in.wav -filter channelmap=1|2|0|5|3|4:5.1 out.wav

Split each channel from an input audio stream into a separate output stream.

It accepts the following parameters:

channel_layout

The channel layout of the input stream. The default is stereo.

channels

A channel layout describing the channels to be extracted as separate output streams or all to extract each input channel as a separate stream. The default is all. Choosing channels not present in channel layout in the input will result in an error.

Examples

• For example, assuming a stereo input MP3 file, ffmpeg -i in.mp3 -filter_complex channelsplit out.mkv will create an output Matroska file with two audio streams, one containing only the left channel and the other the right channel.
• Split a 5.1 WAV file into per-channel files: ffmpeg -i in.wav -filter_complex channelsplit=channel_layout=5.1[FL][FR][FC][LFE][SL][SR] -map [FL] front_left.wav -map [FR] front_right.wav -map [FC] front_center.wav -map [LFE] lfe.wav -map [SL] side_left.wav -map [SR] side_right.wav
• Extract only LFE from a 5.1 WAV file: ffmpeg -i in.wav -filter_complex channelsplit=channel_layout=5.1:channels=LFE[LFE] -map [LFE] lfe.wav

Add a chorus effect to the audio.

Can make a single vocal sound like a chorus, but can also be applied to instrumentation.

Chorus resembles an echo effect with a short delay, but whereas with echo the delay is constant, with chorus, it is varied using using sinusoidal or triangular modulation. The modulation depth defines the range the modulated delay is played before or after the delay. Hence the delayed sound will sound slower or faster, that is the delayed sound tuned around the original one, like in a chorus where some vocals are slightly off key.

It accepts the following parameters:

in_gain

Set input gain. Default is 0.4.

out_gain

Set output gain. Default is 0.4.

delays

Set delays. A typical delay is around 40ms to 60ms.

decays

Set decays.

speeds

Set speeds.

depths

Set depths.

Examples

• A single delay: chorus=0.7:0.9:55:0.4:0.25:2
• Two delays: chorus=0.6:0.9:50|60:0.4|0.32:0.25|0.4:2|1.3
• Fuller sounding chorus with three delays: chorus=0.5:0.9:50|60|40:0.4|0.32|0.3:0.25|0.4|0.3:2|2.3|1.3

Compress or expand the audio’s dynamic range.

It accepts the following parameters:

attacks

decays

A list of times in seconds for each channel over which the instantaneous level of the input signal is averaged to determine its volume. attacks refers to increase of volume and decays refers to decrease of volume. For most situations, the attack time (response to the audio getting louder) should be shorter than the decay time, because the human ear is more sensitive to sudden loud audio than sudden soft audio. A typical value for attack is 0.3 seconds and a typical value for decay is 0.8 seconds. If specified number of attacks & decays is lower than number of channels, the last set attack/decay will be used for all remaining channels.

points

A list of points for the transfer function, specified in dB relative to the maximum possible signal amplitude. Each key points list must be defined using the following syntax: x0/y0|x1/y1|x2/y2|.... or x0/y0 x1/y1 x2/y2 .... The input values must be in strictly increasing order but the transfer function does not have to be monotonically rising. The point 0/0 is assumed but may be overridden (by 0/out-dBn). Typical values for the transfer function are -70/-70|-60/-20|1/0.

soft-knee

Set the curve radius in dB for all joints. It defaults to 0.01.

gain

Set the additional gain in dB to be applied at all points on the transfer function. This allows for easy adjustment of the overall gain. It defaults to 0.

volume

Set an initial volume, in dB, to be assumed for each channel when filtering starts. This permits the user to supply a nominal level initially, so that, for example, a very large gain is not applied to initial signal levels before the companding has begun to operate. A typical value for audio which is initially quiet is -90 dB. It defaults to 0.

delay

Set a delay, in seconds. The input audio is analyzed immediately, but audio is delayed before being fed to the volume adjuster. Specifying a delay approximately equal to the attack/decay times allows the filter to effectively operate in predictive rather than reactive mode. It defaults to 0.

Examples

• Make music with both quiet and loud passages suitable for listening to in a noisy environment: compand=.3|.3:1|1:-90/-60|-60/-40|-40/-30|-20/-20:6:0:-90:0.2 Another example for audio with whisper and explosion parts: compand=0|0:1|1:-90/-900|-70/-70|-30/-9|0/-3:6:0:0:0
• A noise gate for when the noise is at a lower level than the signal: compand=.1|.1:.2|.2:-900/-900|-50.1/-900|-50/-50:.01:0:-90:.1
• Here is another noise gate, this time for when the noise is at a higher level than the signal (making it, in some ways, similar to squelch): compand=.1|.1:.1|.1:-45.1/-45.1|-45/-900|0/-900:.01:45:-90:.1
• 2:1 compression starting at -6dB: compand=points=-80/-80|-6/-6|0/-3.8|20/3.5
• 2:1 compression starting at -9dB: compand=points=-80/-80|-9/-9|0/-5.3|20/2.9
• 2:1 compression starting at -12dB: compand=points=-80/-80|-12/-12|0/-6.8|20/1.9
• 2:1 compression starting at -18dB: compand=points=-80/-80|-18/-18|0/-9.8|20/0.7
• 3:1 compression starting at -15dB: compand=points=-80/-80|-15/-15|0/-10.8|20/-5.2
• Compressor/Gate: compand=points=-80/-105|-62/-80|-15.4/-15.4|0/-12|20/-7.6
• Expander: compand=attacks=0:points=-80/-169|-54/-80|-49.5/-64.6|-41.1/-41.1|-25.8/-15|-10.8/-4.5|0/0|20/8.3
• Hard limiter at -6dB: compand=attacks=0:points=-80/-80|-6/-6|20/-6
• Hard limiter at -12dB: compand=attacks=0:points=-80/-80|-12/-12|20/-12
• Hard noise gate at -35 dB: compand=attacks=0:points=-80/-115|-35.1/-80|-35/-35|20/20
• Soft limiter: compand=attacks=0:points=-80/-80|-12.4/-12.4|-6/-8|0/-6.8|20/-2.8

Compensation Delay Line is a metric based delay to compensate differing positions of microphones or speakers.

For example, you have recorded guitar with two microphones placed in different locations. Because the front of sound wave has fixed speed in normal conditions, the phasing of microphones can vary and depends on their location and interposition. The best sound mix can be achieved when these microphones are in phase (synchronized). Note that a distance of ~30 cm between microphones makes one microphone capture the signal in antiphase to the other microphone. That makes the final mix sound moody. This filter helps to solve phasing problems by adding different delays to each microphone track and make them synchronized.

The best result can be reached when you take one track as base and synchronize other tracks one by one with it. Remember that synchronization/delay tolerance depends on sample rate, too. Higher sample rates will give more tolerance.

The filter accepts the following parameters:

1. Set millimeters distance. This is compensation distance for fine tuning. Default is 0.
2. Set cm distance. This is compensation distance for tightening distance setup. Default is 0.
3. Set meters distance. This is compensation distance for hard distance setup. Default is 0.
4. Set dry amount. Amount of unprocessed (dry) signal. Default is 0.
5. Set wet amount. Amount of processed (wet) signal. Default is 1.
6. Set temperature in degrees Celsius. This is the temperature of the environment. Default is 20.

Apply headphone crossfeed filter.

Crossfeed is the process of blending the left and right channels of stereo audio recording. It is mainly used to reduce extreme stereo separation of low frequencies.

The intent is to produce more speaker like sound to the listener.

The filter accepts the following options:

strength

Set strength of crossfeed. Default is 0.2. Allowed range is from 0 to 1. This sets gain of low shelf filter for side part of stereo image. Default is -6dB. Max allowed is -30db when strength is set to 1.

range

Set soundstage wideness. Default is 0.5. Allowed range is from 0 to 1. This sets cut off frequency of low shelf filter. Default is cut off near 1550 Hz. With range set to 1 cut off frequency is set to 2100 Hz.

slope

Set curve slope of low shelf filter. Default is 0.5. Allowed range is from 0.01 to 1.

level_in

Set input gain. Default is 0.9.

level_out

Set output gain. Default is 1.

Commands

This filter supports the all above options as commands.

Simple algorithm for audio noise sharpening.

This filter linearly increases differences betweeen each audio sample.

The filter accepts the following options:

1. Sets the intensity of effect (default: 2.0). Must be in range between -10.0 to 0 (unchanged sound) to 10.0 (maximum effect). To inverse filtering use negative value.
2. Enable clipping. By default is enabled.

Commands

This filter supports the all above options as commands.

Apply a DC shift to the audio.

This can be useful to remove a DC offset (caused perhaps by a hardware problem in the recording chain) from the audio. The effect of a DC offset is reduced headroom and hence volume. The astats filter can be used to determine if a signal has a DC offset.

shift

Set the DC shift, allowed range is [-1, 1]. It indicates the amount to shift the audio.

limitergain

Optional. It should have a value much less than 1 (e.g. 0.05 or 0.02) and is used to prevent clipping.

Apply de-essing to the audio samples.

1. Set intensity for triggering de-essing. Allowed range is from 0 to 1. Default is 0.
2. Set amount of ducking on treble part of sound. Allowed range is from 0 to 1. Default is 0.5.
3. How much of original frequency content to keep when de-essing. Allowed range is from 0 to 1. Default is 0.5.
4. Set the output mode. It accepts the following values:
   1. Pass input unchanged.
   2. Pass ess filtered out.
   3. Pass only ess. Default value is o.

Measure audio dynamic range.

DR values of 14 and higher is found in very dynamic material. DR of 8 to 13 is found in transition material. And anything less that 8 have very poor dynamics and is very compressed.

The filter accepts the following options:

length

Set window length in seconds used to split audio into segments of equal length. Default is 3 seconds.

Dynamic Audio Normalizer.

This filter applies a certain amount of gain to the input audio in order to bring its peak magnitude to a target level (e.g. 0 dBFS). However, in contrast to more simple normalization algorithms, the Dynamic Audio Normalizer dynamically re-adjusts the gain factor to the input audio. This allows for applying extra gain to the quiet sections of the audio while avoiding distortions or clipping the loud sections. In other words: The Dynamic Audio Normalizer will even out the volume of quiet and loud sections, in the sense that the volume of each section is brought to the same target level. Note, however, that the Dynamic Audio Normalizer achieves this goal without applying dynamic range compressing. It will retain 100% of the dynamic range within each section of the audio file.

framelen, f

Set the frame length in milliseconds. In range from 10 to 8000 milliseconds. Default is 500 milliseconds. The Dynamic Audio Normalizer processes the input audio in small chunks, referred to as frames. This is required, because a peak magnitude has no meaning for just a single sample value. Instead, we need to determine the peak magnitude for a contiguous sequence of sample values. While a standard normalizer would simply use the peak magnitude of the complete file, the Dynamic Audio Normalizer determines the peak magnitude individually for each frame. The length of a frame is specified in milliseconds. By default, the Dynamic Audio Normalizer uses a frame length of 500 milliseconds, which has been found to give good results with most files. Note that the exact frame length, in number of samples, will be determined automatically, based on the sampling rate of the individual input audio file.

gausssize, g

Set the Gaussian filter window size. In range from 3 to 301, must be odd number. Default is 31. Probably the most important parameter of the Dynamic Audio Normalizer is the window size of the Gaussian smoothing filter. The filter’s window size is specified in frames, centered around the current frame. For the sake of simplicity, this must be an odd number. Consequently, the default value of 31 takes into account the current frame, as well as the 15 preceding frames and the 15 subsequent frames. Using a larger window results in a stronger smoothing effect and thus in less gain variation, i.e. slower gain adaptation. Conversely, using a smaller window results in a weaker smoothing effect and thus in more gain variation, i.e. faster gain adaptation. In other words, the more you increase this value, the more the Dynamic Audio Normalizer will behave like a traditional normalization filter. On the contrary, the more you decrease this value, the more the Dynamic Audio Normalizer will behave like a dynamic range compressor.

peak, p

Set the target peak value. This specifies the highest permissible magnitude level for the normalized audio input. This filter will try to approach the target peak magnitude as closely as possible, but at the same time it also makes sure that the normalized signal will never exceed the peak magnitude. A frame’s maximum local gain factor is imposed directly by the target peak magnitude. The default value is 0.95 and thus leaves a headroom of 5%*. It is not recommended to go above this value.

maxgain, m

Set the maximum gain factor. In range from 1.0 to 100.0. Default is 10.0. The Dynamic Audio Normalizer determines the maximum possible (local) gain factor for each input frame, i.e. the maximum gain factor that does not result in clipping or distortion. The maximum gain factor is determined by the frame’s highest magnitude sample. However, the Dynamic Audio Normalizer additionally bounds the frame’s maximum gain factor by a predetermined (global) maximum gain factor. This is done in order to avoid excessive gain factors in silent or almost silent frames. By default, the maximum gain factor is 10.0, For most inputs the default value should be sufficient and it usually is not recommended to increase this value. Though, for input with an extremely low overall volume level, it may be necessary to allow even higher gain factors. Note, however, that the Dynamic Audio Normalizer does not simply apply a hard threshold (i.e. cut off values above the threshold). Instead, a sigmoid threshold function will be applied. This way, the gain factors will smoothly approach the threshold value, but never exceed that value.

targetrms, r

Set the target RMS. In range from 0.0 to 1.0. Default is 0.0 - disabled. By default, the Dynamic Audio Normalizer performs peak normalization. This means that the maximum local gain factor for each frame is defined (only) by the frame’s highest magnitude sample. This way, the samples can be amplified as much as possible without exceeding the maximum signal level, i.e. without clipping. Optionally, however, the Dynamic Audio Normalizer can also take into account the frame’s root mean square, abbreviated RMS. In electrical engineering, the RMS is commonly used to determine the power of a time-varying signal. It is therefore considered that the RMS is a better approximation of the perceived loudness than just looking at the signal’s peak magnitude. Consequently, by adjusting all frames to a constant RMS value, a uniform perceived loudness can be established. If a target RMS value has been specified, a frame’s local gain factor is defined as the factor that would result in exactly that RMS value. Note, however, that the maximum local gain factor is still restricted by the frame’s highest magnitude sample, in order to prevent clipping.

coupling, n

Enable channels coupling. By default is enabled. By default, the Dynamic Audio Normalizer will amplify all channels by the same amount. This means the same gain factor will be applied to all channels, i.e. the maximum possible gain factor is determined by the loudest channel. However, in some recordings, it may happen that the volume of the different channels is uneven, e.g. one channel may be quieter than the other one(s). In this case, this option can be used to disable the channel coupling. This way, the gain factor will be determined independently for each channel, depending only on the individual channel’s highest magnitude sample. This allows for harmonizing the volume of the different channels.

correctdc, c

Enable DC bias correction. By default is disabled. An audio signal (in the time domain) is a sequence of sample values. In the Dynamic Audio Normalizer these sample values are represented in the -1.0 to 1.0 range, regardless of the original input format. Normally, the audio signal, or waveform, should be centered around the zero point. That means if we calculate the mean value of all samples in a file, or in a single frame, then the result should be 0.0 or at least very close to that value. If, however, there is a significant deviation of the mean value from 0.0, in either positive or negative direction, this is referred to as a DC bias or DC offset. Since a DC bias is clearly undesirable, the Dynamic Audio Normalizer provides optional DC bias correction. With DC bias correction enabled, the Dynamic Audio Normalizer will determine the mean value, or DC correction offset, of each input frame and subtract that value from all of the frame’s sample values which ensures those samples are centered around 0.0 again. Also, in order to avoid gaps at the frame boundaries, the DC correction offset values will be interpolated smoothly between neighbouring frames.

altboundary, b

Enable alternative boundary mode. By default is disabled. The Dynamic Audio Normalizer takes into account a certain neighbourhood around each frame. This includes the preceding frames as well as the subsequent frames. However, for the boundary frames, located at the very beginning and at the very end of the audio file, not all neighbouring frames are available. In particular, for the first few frames in the audio file, the preceding frames are not known. And, similarly, for the last few frames in the audio file, the subsequent frames are not known. Thus, the question arises which gain factors should be assumed for the missing frames in the boundary region. The Dynamic Audio Normalizer implements two modes to deal with this situation. The default boundary mode assumes a gain factor of exactly 1.0 for the missing frames, resulting in a smooth fade in and fade out at the beginning and at the end of the input, respectively.

compress, s

Set the compress factor. In range from 0.0 to 30.0. Default is 0.0. By default, the Dynamic Audio Normalizer does not apply traditional compression. This means that signal peaks will not be pruned and thus the full dynamic range will be retained within each local neighbourhood. However, in some cases it may be desirable to combine the Dynamic Audio Normalizer’s normalization algorithm with a more traditional compression. For this purpose, the Dynamic Audio Normalizer provides an optional compression (thresholding) function. If (and only if) the compression feature is enabled, all input frames will be processed by a soft knee thresholding function prior to the actual normalization process. Put simply, the thresholding function is going to prune all samples whose magnitude exceeds a certain threshold value. However, the Dynamic Audio Normalizer does not simply apply a fixed threshold value. Instead, the threshold value will be adjusted for each individual frame. In general, smaller parameters result in stronger compression, and vice versa. Values below 3.0 are not recommended, because audible distortion may appear.

threshold, t

Set the target threshold value. This specifies the lowest permissible magnitude level for the audio input which will be normalized. If input frame volume is above this value frame will be normalized. Otherwise frame may not be normalized at all. The default value is set to 0, which means all input frames will be normalized. This option is mostly useful if digital noise is not wanted to be amplified.

Commands

This filter supports the all above options as commands.

Make audio easier to listen to on headphones.

This filter adds `cues’ to 44.1kHz stereo (i.e. audio CD format) audio so that when listened to on headphones the stereo image is moved from inside your head (standard for headphones) to outside and in front of the listener (standard for speakers).

Ported from SoX.

Apply a two-pole peaking equalisation (EQ) filter. With this filter, the signal-level at and around a selected frequency can be increased or decreased, whilst (unlike bandpass and bandreject filters) that at all other frequencies is unchanged.

In order to produce complex equalisation curves, this filter can be given several times, each with a different central frequency.

The filter accepts the following options:

frequency, f

Set the filter’s central frequency in Hz.

width_type, t

Set method to specify band-width of filter.

1. Hz
2. Q-Factor
3. octave
4. slope
5. kHz

width, w

Specify the band-width of a filter in width_type units.

gain, g

Set the required gain or attenuation in dB. Beware of clipping when using a positive gain.

mix, m

How much to use filtered signal in output. Default is 1. Range is between 0 and 1.

channels, c

Specify which channels to filter, by default all available are filtered.

normalize, n

Normalize biquad coefficients, by default is disabled. Enabling it will normalize magnitude response at DC to 0dB.

transform, a

Set transform type of IIR filter.

precision, r

Set precison of filtering.

auto

Pick automatic sample format depending on surround filters.

s16

Always use signed 16-bit.

s32

Always use signed 32-bit.

f32

Always use float 32-bit.

f64

Always use float 64-bit.

nil

Examples

• Attenuate 10 dB at 1000 Hz, with a bandwidth of 200 Hz: equalizer=f=1000:t=h:width=200:g=-10
• Apply 2 dB gain at 1000 Hz with Q 1 and attenuate 5 dB at 100 Hz with Q 2: equalizer=f=1000:t=q:w=1:g=2,equalizer=f=100:t=q:w=2:g=-5

Commands

This filter supports the following commands:

frequency, f

Change equalizer frequency. Syntax for the command is

"/frequency/"

width_type, t

Change equalizer width_type. Syntax for the command is : “width_type”

width, w

Change equalizer width. Syntax for the command is : “width”

gain, g

Change equalizer gain. Syntax for the command is : “gain”

mix, m

Change equalizer mix. Syntax for the command is : “mix”

Linearly increases the difference between left and right channels which adds some sort of live effect to playback.

The filter accepts the following options:

1. Sets the difference coefficient (default: 2.5). 0.0 means mono sound (average of both channels), with 1.0 sound will be unchanged, with -1.0 left and right channels will be swapped.
2. Enable clipping. By default is enabled.

Commands

This filter supports the all above options as commands.

Apply FIR Equalization using arbitrary frequency response.

The filter accepts the following option:

gain

Set gain curve equation (in dB). The expression can contain variables:

1. the evaluated frequency
2. sample rate
3. channel number, set to 0 when multichannels evaluation is disabled
4. channel id, see libavutil/channel_layout.h, set to the first channel id when multichannels evaluation is disabled
5. number of channels
6. channel_layout, see libavutil/channel_layout.h

and functions:

gain_interpolate(f)

interpolate gain on frequency f based on gain_entry

cubic_interpolate(f)

same as gain_interpolate, but smoother

This option is also available as command. Default is gain_interpolate(f).

gain_entry

Set gain entry for gain_interpolate function. The expression can contain functions:

entry(f, g)

store gain entry at frequency f with value g

This option is also available as command.

delay

Set filter delay in seconds. Higher value means more accurate. Default is 0.01.

accuracy

Set filter accuracy in Hz. Lower value means more accurate. Default is 5.

wfunc

Set window function. Acceptable values are:

rectangular

rectangular window, useful when gain curve is already smooth

hann

hann window (default)

hamming

hamming window

blackman

blackman window

nuttall3

3-terms continuous 1st derivative nuttall window

mnuttall3

minimum 3-terms discontinuous nuttall window

nuttall

4-terms continuous 1st derivative nuttall window

bnuttall

minimum 4-terms discontinuous nuttall (blackman-nuttall) window

bharris

blackman-harris window

tukey

tukey window

fixed

If enabled, use fixed number of audio samples. This improves speed when filtering with large delay. Default is disabled.

multi

Enable multichannels evaluation on gain. Default is disabled.

zero_phase

Enable zero phase mode by subtracting timestamp to compensate delay. Default is disabled.

scale

Set scale used by gain. Acceptable values are:

linlin

linear frequency, linear gain

linlog

linear frequency, logarithmic (in dB) gain (default)

loglin

logarithmic (in octave scale where 20 Hz is 0) frequency, linear gain

loglog

logarithmic frequency, logarithmic gain

dumpfile

Set file for dumping, suitable for gnuplot.

dumpscale

Set scale for dumpfile. Acceptable values are same with scale option. Default is linlog.

fft2

Enable 2-channel convolution using complex FFT. This improves speed significantly. Default is disabled.

min_phase

Enable minimum phase impulse response. Default is disabled.

Examples

• lowpass at 1000 Hz: firequalizer=gain=if(lt(f,1000), 0, -INF)
• lowpass at 1000 Hz with gain_entry: firequalizer=gain_entry=entry(1000,0); entry(1001, -INF)
• custom equalization: firequalizer=gain_entry=entry(100,0); entry(400, -4); entry(1000, -6); entry(2000, 0)
• higher delay with zero phase to compensate delay: firequalizer=delay=0.1:fixed=on:zero_phase=on
• lowpass on left channel, highpass on right channel: firequalizer=gain=if(eq(chid,1), gain_interpolate(f), if(eq(chid,2), gain_interpolate(1e6+f), 0)) :gain_entry=entry(1000, 0); entry(1001,-INF); entry(1e6+1000,0):multi=on

Apply a flanging effect to the audio.

The filter accepts the following options:

delay

Set base delay in milliseconds. Range from 0 to 30. Default value is 0.

depth

Set added sweep delay in milliseconds. Range from 0 to 10. Default value is 2.

regen

Set percentage regeneration (delayed signal feedback). Range from -95 to 95. Default value is 0.

width

Set percentage of delayed signal mixed with original. Range from 0 to 100. Default value is 71.

speed

Set sweeps per second (Hz). Range from 0.1 to 10. Default value is 0.5.

shape

Set swept wave shape, can be triangular or sinusoidal. Default value is sinusoidal.

phase

Set swept wave percentage-shift for multi channel. Range from 0 to 100. Default value is 25.

interp

Set delay-line interpolation, linear or quadratic. Default is linear.

Apply Haas effect to audio.

Note that this makes most sense to apply on mono signals. With this filter applied to mono signals it give some directionality and stretches its stereo image.

The filter accepts the following options:

level_in

Set input level. By default is 1, or 0dB

level_out

Set output level. By default is 1, or 0dB.

side_gain

Set gain applied to side part of signal. By default is 1.

middle_source

Set kind of middle source. Can be one of the following:

left

Pick left channel.

right

Pick right channel.

mid

Pick middle part signal of stereo image.

side

Pick side part signal of stereo image.

middle_phase

Change middle phase. By default is disabled.

left_delay

Set left channel delay. By default is 2.05 milliseconds.

left_balance

Set left channel balance. By default is -1.

left_gain

Set left channel gain. By default is 1.

left_phase

Change left phase. By default is disabled.

right_delay

Set right channel delay. By defaults is 2.12 milliseconds.

right_balance

Set right channel balance. By default is 1.

right_gain

Set right channel gain. By default is 1.

right_phase

Change right phase. By default is enabled.

Decodes High Definition Compatible Digital (HDCD) data. A 16-bit PCM stream with embedded HDCD codes is expanded into a 20-bit PCM stream.

The filter supports the Peak Extend and Low-level Gain Adjustment features of HDCD, and detects the Transient Filter flag.

ffmpeg -i HDCD16.flac -af hdcd OUT24.flac

When using the filter with wav, note the default encoding for wav is 16-bit, so the resulting 20-bit stream will be truncated back to 16-bit. Use something like -acodec pcm_s24le after the filter to get 24-bit PCM output.

ffmpeg -i HDCD16.wav -af hdcd OUT16.wav ffmpeg -i HDCD16.wav -af hdcd -c:a pcm_s24le OUT24.wav

The filter accepts the following options:

disable_autoconvert

Disable any automatic format conversion or resampling in the filter graph.

process_stereo

Process the stereo channels together. If target_gain does not match between channels, consider it invalid and use the last valid target_gain.

cdt_ms

Set the code detect timer period in ms.

force_pe

Always extend peaks above -3dBFS even if PE isn’t signaled.

analyze_mode

Replace audio with a solid tone and adjust the amplitude to signal some specific aspect of the decoding process. The output file can be loaded in an audio editor alongside the original to aid analysis. analyze_mode=pe:force_pe=true can be used to see all samples above the PE level. Modes are:

0, off

Disabled

1, lle

Gain adjustment level at each sample

2, pe

Samples where peak extend occurs

3, cdt

Samples where the code detect timer is active

4, tgm

Samples where the target gain does not match between channels

Apply head-related transfer functions (HRTFs) to create virtual loudspeakers around the user for binaural listening via headphones. The HRIRs are provided via additional streams, for each channel one stereo input stream is needed.

The filter accepts the following options:

map

Set mapping of input streams for convolution. The argument is a ’|’-separated list of channel names in order as they are given as additional stream inputs for filter. This also specify number of input streams. Number of input streams must be not less than number of channels in first stream plus one.

gain

Set gain applied to audio. Value is in dB. Default is 0.

type

Set processing type. Can be time or freq. time is processing audio in time domain which is slow. freq is processing audio in frequency domain which is fast. Default is freq.

lfe

Set custom gain for LFE channels. Value is in dB. Default is 0.

size

Set size of frame in number of samples which will be processed at once. Default value is 1024. Allowed range is from 1024 to 96000.

hrir

Set format of hrir stream. Default value is stereo. Alternative value is multich. If value is set to stereo, number of additional streams should be greater or equal to number of input channels in first input stream. Also each additional stream should have stereo number of channels. If value is set to multich, number of additional streams should be exactly one. Also number of input channels of additional stream should be equal or greater than twice number of channels of first input stream.

Examples

• Full example using wav files as coefficients with amovie filters for 7.1 downmix, each amovie filter use stereo file with IR coefficients as input. The files give coefficients for each position of virtual loudspeaker: ffmpeg -i input.wav -filter_complex “amovie=azi_270_ele_0_DFC.wav[sr];amovie=azi_90_ele_0_DFC.wav[sl];amovie=azi_225_ele_0_DFC.wav[br];amovie=azi_135_ele_0_DFC.wav[bl];amovie=azi_0_ele_0_DFC.wav,asplit[fc][lfe];amovie=azi_35_ele_0_DFC.wav[fl];amovie=azi_325_ele_0_DFC.wav[fr];[0:a][fl][fr][fc][lfe][bl][br][sl][sr]headphone=FL|FR|FC|LFE|BL|BR|SL|SR” output.wav
• Full example using wav files as coefficients with amovie filters for 7.1 downmix, but now in multich hrir format. ffmpeg -i input.wav -filter_complex “amovie=minp.wav[hrirs];[0:a][hrirs]headphone=map=FL|FR|FC|LFE|BL|BR|SL|SR:hrir=multich” output.wav

Apply a high-pass filter with 3dB point frequency. The filter can be either single-pole, or double-pole (the default). The filter roll off at 6dB per pole per octave (20dB per pole per decade).

The filter accepts the following options:

frequency, f

Set frequency in Hz. Default is 3000.

poles, p

Set number of poles. Default is 2.

width_type, t

Set method to specify band-width of filter.

1. Hz
2. Q-Factor
3. octave
4. slope
5. kHz

width, w

Specify the band-width of a filter in width_type units. Applies only to double-pole filter. The default is 0.707q and gives a Butterworth response.

mix, m

How much to use filtered signal in output. Default is 1. Range is between 0 and 1.

channels, c

Specify which channels to filter, by default all available are filtered.

normalize, n

Normalize biquad coefficients, by default is disabled. Enabling it will normalize magnitude response at DC to 0dB.

transform, a

Set transform type of IIR filter.

precision, r

Set precison of filtering.

auto

Pick automatic sample format depending on surround filters.

s16

Always use signed 16-bit.

s32

Always use signed 32-bit.

f32

Always use float 32-bit.

f64

Always use float 64-bit.

nil

Commands

This filter supports the following commands:

frequency, f

Change highpass frequency. Syntax for the command is : “frequency”

width_type, t

Change highpass width_type. Syntax for the command is

"/width_type/"

width, w

Change highpass width. Syntax for the command is : “width”

mix, m

Change highpass mix. Syntax for the command is : “mix”

Join multiple input streams into one multi-channel stream.

It accepts the following parameters:

inputs

The number of input streams. It defaults to 2.

channel_layout

The desired output channel layout. It defaults to stereo.

map

Map channels from inputs to output. The argument is a ’|’-separated list of mappings, each in the input_idx=/.=/=in_channel=/=-/=out_channel form. input_idx is the 0-based index of the input stream. in_channel can be either the name of the input channel (e.g. FL for front left) or its index in the specified input stream. out_channel is the name of the output channel.

The filter will attempt to guess the mappings when they are not specified explicitly. It does so by first trying to find an unused matching input channel and if that fails it picks the first unused input channel.

Join 3 inputs (with properly set channel layouts):

ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex join=inputs=3 OUTPUT

Build a 5.1 output from 6 single-channel streams:

ffmpeg -i fl -i fr -i fc -i sl -i sr -i lfe -filter_complex join=inputs=6:channel_layout=5.1:map=0.0-FL|1.0-FR|2.0-FC|3.0-SL|4.0-SR|5.0-LFE out

Load a LADSPA (Linux Audio Developer’s Simple Plugin API) plugin.

To enable compilation of this filter you need to configure FFmpeg with --enable-ladspa.

file, f

Specifies the name of LADSPA plugin library to load. If the environment variable LADSPA_PATH is defined, the LADSPA plugin is searched in each one of the directories specified by the colon separated list in LADSPA_PATH, otherwise in the standard LADSPA paths, which are in this order: HOME.ladspa/lib//, usr/local/lib/ladspa, usr/lib/ladspa.

plugin, p

Specifies the plugin within the library. Some libraries contain only one plugin, but others contain many of them. If this is not set filter will list all available plugins within the specified library.

controls, c

Set the ’|’ separated list of controls which are zero or more floating point values that determine the behavior of the loaded plugin (for example delay, threshold or gain). Controls need to be defined using the following syntax: c0=/value0/|c1=/value1/|c2=/value2/|…, where valuei is the value set on the i-th control. Alternatively they can be also defined using the following syntax: value0/|/value1/|/value2/|…, where /valuei is the value set on the i-th control. If controls is set to help, all available controls and their valid ranges are printed.

sample_rate, s

Specify the sample rate, default to 44100. Only used if plugin have zero inputs.

nb_samples, n

Set the number of samples per channel per each output frame, default is 1024. Only used if plugin have zero inputs.

duration, d

Set the minimum duration of the sourced audio. See the Time duration section in the ffmpeg-utils (1) manual for the accepted syntax. Note that the resulting duration may be greater than the specified duration, as the generated audio is always cut at the end of a complete frame. If not specified, or the expressed duration is negative, the audio is supposed to be generated forever. Only used if plugin have zero inputs.

latency, l

Enable latency compensation, by default is disabled. Only used if plugin have inputs.

Examples

• List all available plugins within amp (LADSPA example plugin) library: ladspa=file=amp
• List all available controls and their valid ranges for vcf_notch plugin from VCF library: ladspa=f=vcf:p=vcf_notch:c=help
• Simulate low quality audio equipment using Computer Music Toolkit (CMT) plugin library: ladspa=file=cmt:plugin=lofi:controls=c0=22|c1=12|c2=12
• Add reverberation to the audio using TAP-plugins (Tom’s Audio Processing plugins): ladspa=file=tap_reverb:tap_reverb
• Generate white noise, with 0.2 amplitude: ladspa=file=cmt:noise_source_white:c=c0=.2
• Generate 20 bpm clicks using plugin C* Click - Metronome from the C* Audio Plugin Suite (CAPS) library: ladspa=file=caps:Click:c=c1=20
• Apply C* Eq10X2 - Stereo 10-band equaliser effect: ladspa=caps:Eq10X2:c=c0=-48|c9=-24|c3=12|c4=2
• Increase volume by 20dB using fast lookahead limiter from Steve Harris SWH Plugins collection: ladspa=fast_lookahead_limiter_1913:fastLookaheadLimiter:20|0|2
• Attenuate low frequencies using Multiband EQ from Steve Harris SWH Plugins collection: ladspa=mbeq_1197:mbeq:-24|-24|-24|0|0|0|0|0|0|0|0|0|0|0|0
• Reduce stereo image using Narrower from the C* Audio Plugin Suite (CAPS) library: ladspa=caps:Narrower
• Another white noise, now using C* Audio Plugin Suite (CAPS) library: ladspa=caps:White:.2
• Some fractal noise, using C* Audio Plugin Suite (CAPS) library: ladspa=caps:Fractal:c=c1=1
• Dynamic volume normalization using VLevel plugin: ladspa=vlevel-ladspa:vlevel_mono

Commands

This filter supports the following commands:

cN

Modify the N-th control value. If the specified value is not valid, it is ignored and prior one is kept.

EBU R128 loudness normalization. Includes both dynamic and linear normalization modes. Support for both single pass (livestreams, files) and double pass (files) modes. This algorithm can target IL, LRA, and maximum true peak. In dynamic mode, to accurately detect true peaks, the audio stream will be upsampled to 192 kHz. Use the -ar option or aresample filter to explicitly set an output sample rate.

The filter accepts the following options:

I, i

Set integrated loudness target. Range is -70.0 - -5.0. Default value is -24.0.

LRA, lra

Set loudness range target. Range is 1.0 - 20.0. Default value is 7.0.

TP, tp

Set maximum true peak. Range is -9.0 - +0.0. Default value is -2.0.

measured_I, measured_i

Measured IL of input file. Range is -99.0 - +0.0.

measured_LRA, measured_lra

Measured LRA of input file. Range is 0.0 - 99.0.

measured_TP, measured_tp

Measured true peak of input file. Range is -99.0 - +99.0.

measured_thresh

Measured threshold of input file. Range is -99.0 - +0.0.

offset

Set offset gain. Gain is applied before the true-peak limiter. Range is -99.0 - +99.0. Default is +0.0.

linear

Normalize by linearly scaling the source audio. measured_I, measured_LRA, measured_TP, and measured_thresh must all be specified. Target LRA shouldn’t be lower than source LRA and the change in integrated loudness shouldn’t result in a true peak which exceeds the target TP. If any of these conditions aren’t met, normalization mode will revert to dynamic. Options are true or false. Default is true.

dual_mono

Treat mono input files as dual-mono. If a mono file is intended for playback on a stereo system, its EBU R128 measurement will be perceptually incorrect. If set to true, this option will compensate for this effect. Multi-channel input files are not affected by this option. Options are true or false. Default is false.

print_format

Set print format for stats. Options are summary, json, or none. Default value is none.

Apply a low-pass filter with 3dB point frequency. The filter can be either single-pole or double-pole (the default). The filter roll off at 6dB per pole per octave (20dB per pole per decade).

The filter accepts the following options:

frequency, f

Set frequency in Hz. Default is 500.

poles, p

Set number of poles. Default is 2.

width_type, t

Set method to specify band-width of filter.

1. Hz
2. Q-Factor
3. octave
4. slope
5. kHz

width, w

Specify the band-width of a filter in width_type units. Applies only to double-pole filter. The default is 0.707q and gives a Butterworth response.

mix, m

How much to use filtered signal in output. Default is 1. Range is between 0 and 1.

channels, c

Specify which channels to filter, by default all available are filtered.

normalize, n

Normalize biquad coefficients, by default is disabled. Enabling it will normalize magnitude response at DC to 0dB.

transform, a

Set transform type of IIR filter.

precision, r

Set precison of filtering.

auto

Pick automatic sample format depending on surround filters.

s16

Always use signed 16-bit.

s32

Always use signed 32-bit.

f32

Always use float 32-bit.

f64

Always use float 64-bit.

nil

Examples

• Lowpass only LFE channel, it LFE is not present it does nothing: lowpass=c=LFE

Commands

This filter supports the following commands:

frequency, f

Change lowpass frequency. Syntax for the command is : “frequency”

width_type, t

Change lowpass width_type. Syntax for the command is

"/width_type/"

width, w

Change lowpass width. Syntax for the command is : “width”

mix, m

Change lowpass mix. Syntax for the command is : “mix”

Load a LV2 (LADSPA Version 2) plugin.

To enable compilation of this filter you need to configure FFmpeg with --enable-lv2.

plugin, p

Specifies the plugin URI. You may need to escape ’:’.

controls, c

Set the ’|’ separated list of controls which are zero or more floating point values that determine the behavior of the loaded plugin (for example delay, threshold or gain). If controls is set to help, all available controls and their valid ranges are printed.

sample_rate, s

Specify the sample rate, default to 44100. Only used if plugin have zero inputs.

nb_samples, n

Set the number of samples per channel per each output frame, default is 1024. Only used if plugin have zero inputs.

duration, d

Set the minimum duration of the sourced audio. See the Time duration section in the ffmpeg-utils (1) manual for the accepted syntax. Note that the resulting duration may be greater than the specified duration, as the generated audio is always cut at the end of a complete frame. If not specified, or the expressed duration is negative, the audio is supposed to be generated forever. Only used if plugin have zero inputs.

Examples

• Apply bass enhancer plugin from Calf: lv2=p=http\\\\://calf.sourceforge.net/plugins/BassEnhancer:c=amount=2
• Apply vinyl plugin from Calf: lv2=p=http\\\\://calf.sourceforge.net/plugins/Vinyl:c=drone=0.2|aging=0.5
• Apply bit crusher plugin from ArtyFX: lv2=p=http\\\\://www.openavproductions.com/artyfx#bitta:c=crush=0.3

Multiband Compress or expand the audio’s dynamic range.

The input audio is divided into bands using 4th order Linkwitz-Riley IIRs. This is akin to the crossover of a loudspeaker, and results in flat frequency response when absent compander action.

It accepts the following parameters:

args

This option syntax is: attack,decay,[attack,decay..] soft-knee points crossover_frequency [delay [initial_volume [gain]]] | attack,decay … For explanation of each item refer to compand filter documentation.

Mix channels with specific gain levels. The filter accepts the output channel layout followed by a set of channels definitions.

This filter is also designed to efficiently remap the channels of an audio stream.

The filter accepts parameters of the form: “/l/|/outdef/|/outdef/|…”

1. output channel layout or number of channels
2. output channel specification, of the form: “out_name/=[/gain/*]/in_name[(+-)[/gain/*]/in_name/…]”
3. output channel to define, either a channel name (FL, FR, etc.) or a channel number (c0, c1, etc.)
4. multiplicative coefficient for the channel, 1 leaving the volume unchanged
5. input channel to use, see out_name for details; it is not possible to mix named and numbered input channels

If the `=’ in a channel specification is replaced by `<’, then the gains for that specification will be renormalized so that the total is 1, thus avoiding clipping noise.

Mixing examples

For example, if you want to down-mix from stereo to mono, but with a bigger factor for the left channel:

pan=1c|c0=0.9*c0+0.1*c1

A customized down-mix to stereo that works automatically for 3-, 4-, 5- and 7-channels surround:

pan=stereo| FL < FL + 0.5*FC + 0.6*BL + 0.6*SL | FR < FR + 0.5*FC + 0.6*BR + 0.6*SR

Note that ffmpeg integrates a default down-mix (and up-mix) system that should be preferred (see -ac option) unless you have very specific needs.

Remapping examples

The channel remapping will be effective if, and only if:

*<gain coefficients are zeroes or ones,>

*<only one input per channel output,>

If all these conditions are satisfied, the filter will notify the user (Pure channel mapping detected), and use an optimized and lossless method to do the remapping.

For example, if you have a 5.1 source and want a stereo audio stream by dropping the extra channels:

pan=“stereo| c0=FL | c1=FR”

Given the same source, you can also switch front left and front right channels and keep the input channel layout:

pan=“5.1| c0=c1 | c1=c0 | c2=c2 | c3=c3 | c4=c4 | c5=c5”

If the input is a stereo audio stream, you can mute the front left channel (and still keep the stereo channel layout) with:

pan=“stereo|c1=c1”

Still with a stereo audio stream input, you can copy the right channel in both front left and right:

pan=“stereo| c0=FR | c1=FR”

ReplayGain scanner filter. This filter takes an audio stream as an input and outputs it unchanged. At end of filtering it displays track_gain and track_peak.

Convert the audio sample format, sample rate and channel layout. It is not meant to be used directly.

Apply time-stretching and pitch-shifting with librubberband.

To enable compilation of this filter, you need to configure FFmpeg with --enable-librubberband.

The filter accepts the following options:

tempo

Set tempo scale factor.

pitch

Set pitch scale factor.

transients

Set transients detector. Possible values are:

crisp

mixed

smooth

detector

Set detector. Possible values are:

compound

percussive

soft

nil

phase

Set phase. Possible values are:

laminar

independent

nil

window

Set processing window size. Possible values are:

standard

short

long

nil

smoothing

Set smoothing. Possible values are:

off

on

nil

formant

Enable formant preservation when shift pitching. Possible values are:

shifted

preserved

nil

pitchq

Set pitch quality. Possible values are:

quality

speed

consistency

nil

channels

Set channels. Possible values are:

apart

together

nil

Commands

This filter supports the following commands:

tempo

Change filter tempo scale factor. Syntax for the command is : “tempo”

pitch

Change filter pitch scale factor. Syntax for the command is : “pitch”

This filter acts like normal compressor but has the ability to compress detected signal using second input signal. It needs two input streams and returns one output stream. First input stream will be processed depending on second stream signal. The filtered signal then can be filtered with other filters in later stages of processing. See pan and amerge filter.

The filter accepts the following options:

level_in

Set input gain. Default is 1. Range is between 0.015625 and 64.

mode

Set mode of compressor operation. Can be upward or downward. Default is downward.

threshold

If a signal of second stream raises above this level it will affect the gain reduction of first stream. By default is 0.125. Range is between 0.00097563 and 1.

ratio

Set a ratio about which the signal is reduced. 1:2 means that if the level raised 4dB above the threshold, it will be only 2dB above after the reduction. Default is 2. Range is between 1 and 20.

attack

Amount of milliseconds the signal has to rise above the threshold before gain reduction starts. Default is 20. Range is between 0.01 and 2000.

release

Amount of milliseconds the signal has to fall below the threshold before reduction is decreased again. Default is 250. Range is between 0.01 and 9000.

makeup

Set the amount by how much signal will be amplified after processing. Default is 1. Range is from 1 to 64.

knee

Curve the sharp knee around the threshold to enter gain reduction more softly. Default is 2.82843. Range is between 1 and 8.

link

Choose if the average level between all channels of side-chain stream or the louder(maximum) channel of side-chain stream affects the reduction. Default is average.

detection

Should the exact signal be taken in case of peak or an RMS one in case of rms. Default is rms which is mainly smoother.

level_sc

Set sidechain gain. Default is 1. Range is between 0.015625 and 64.

(no term)

How much to use compressed signal in output. Default is 1. Range is between 0 and 1.

Commands

This filter supports the all above options as commands.

Examples

• Full ffmpeg example taking 2 audio inputs, 1st input to be compressed depending on the signal of 2nd input and later compressed signal to be merged with 2nd input: ffmpeg -i main.flac -i sidechain.flac -filter_complex “[1:a]asplit=2[sc][mix];[0:a][sc]sidechaincompress[compr];[compr][mix]amerge”

A sidechain gate acts like a normal (wideband) gate but has the ability to filter the detected signal before sending it to the gain reduction stage. Normally a gate uses the full range signal to detect a level above the threshold. For example: If you cut all lower frequencies from your sidechain signal the gate will decrease the volume of your track only if not enough highs appear. With this technique you are able to reduce the resonation of a natural drum or remove rumbling of muted strokes from a heavily distorted guitar. It needs two input streams and returns one output stream. First input stream will be processed depending on second stream signal.

The filter accepts the following options:

level_in

Set input level before filtering. Default is 1. Allowed range is from 0.015625 to 64.

mode

Set the mode of operation. Can be upward or downward. Default is downward. If set to upward mode, higher parts of signal will be amplified, expanding dynamic range in upward direction. Otherwise, in case of downward lower parts of signal will be reduced.

range

Set the level of gain reduction when the signal is below the threshold. Default is 0.06125. Allowed range is from 0 to 1. Setting this to 0 disables reduction and then filter behaves like expander.

threshold

If a signal rises above this level the gain reduction is released. Default is 0.125. Allowed range is from 0 to 1.

ratio

Set a ratio about which the signal is reduced. Default is 2. Allowed range is from 1 to 9000.

attack

Amount of milliseconds the signal has to rise above the threshold before gain reduction stops. Default is 20 milliseconds. Allowed range is from 0.01 to 9000.

release

Amount of milliseconds the signal has to fall below the threshold before the reduction is increased again. Default is 250 milliseconds. Allowed range is from 0.01 to 9000.

makeup

Set amount of amplification of signal after processing. Default is 1. Allowed range is from 1 to 64.

knee

Curve the sharp knee around the threshold to enter gain reduction more softly. Default is 2.828427125. Allowed range is from 1 to 8.

detection

Choose if exact signal should be taken for detection or an RMS like one. Default is rms. Can be peak or rms.

link

Choose if the average level between all channels or the louder channel affects the reduction. Default is average. Can be average or maximum.

level_sc

Set sidechain gain. Default is 1. Range is from 0.015625 to 64.

Commands

This filter supports the all above options as commands.

Detect silence in an audio stream.

This filter logs a message when it detects that the input audio volume is less or equal to a noise tolerance value for a duration greater or equal to the minimum detected noise duration.

The printed times and duration are expressed in seconds. The lavfi.silence_start or lavfi.silence_start.X metadata key is set on the first frame whose timestamp equals or exceeds the detection duration and it contains the timestamp of the first frame of the silence.

The lavfi.silence_duration or lavfi.silence_duration.X and lavfi.silence_end or lavfi.silence_end.X metadata keys are set on the first frame after the silence. If mono is enabled, and each channel is evaluated separately, the .X suffixed keys are used, and X corresponds to the channel number.

The filter accepts the following options:

noise, n

Set noise tolerance. Can be specified in dB (in case dB is appended to the specified value) or amplitude ratio. Default is -60dB, or 0.001.

duration, d

Set silence duration until notification (default is 2 seconds). See the Time duration section in the ffmpeg-utils (1) manual for the accepted syntax.

mono, m

Process each channel separately, instead of combined. By default is disabled.

Examples

• Detect 5 seconds of silence with -50dB noise tolerance: silencedetect=n=-50dB:d=5
• Complete example with ffmpeg to detect silence with 0.0001 noise tolerance in silence.mp3: ffmpeg -i silence.mp3 -af silencedetect=noise=0.0001 -f null -

Remove silence from the beginning, middle or end of the audio.

The filter accepts the following options:

start_periods

This value is used to indicate if audio should be trimmed at beginning of the audio. A value of zero indicates no silence should be trimmed from the beginning. When specifying a non-zero value, it trims audio up until it finds non-silence. Normally, when trimming silence from beginning of audio the start_periods will be 1 but it can be increased to higher values to trim all audio up to specific count of non-silence periods. Default value is 0.

start_duration

Specify the amount of time that non-silence must be detected before it stops trimming audio. By increasing the duration, bursts of noises can be treated as silence and trimmed off. Default value is 0.

start_threshold

This indicates what sample value should be treated as silence. For digital audio, a value of 0 may be fine but for audio recorded from analog, you may wish to increase the value to account for background noise. Can be specified in dB (in case dB is appended to the specified value) or amplitude ratio. Default value is 0.

start_silence

Specify max duration of silence at beginning that will be kept after trimming. Default is 0, which is equal to trimming all samples detected as silence.

start_mode

Specify mode of detection of silence end in start of multi-channel audio. Can be any or all. Default is any. With any, any sample that is detected as non-silence will cause stopped trimming of silence. With all, only if all channels are detected as non-silence will cause stopped trimming of silence.

stop_periods

Set the count for trimming silence from the end of audio. To remove silence from the middle of a file, specify a stop_periods that is negative. This value is then treated as a positive value and is used to indicate the effect should restart processing as specified by start_periods, making it suitable for removing periods of silence in the middle of the audio. Default value is 0.

stop_duration

Specify a duration of silence that must exist before audio is not copied any more. By specifying a higher duration, silence that is wanted can be left in the audio. Default value is 0.

stop_threshold

This is the same as start_threshold but for trimming silence from the end of audio. Can be specified in dB (in case dB is appended to the specified value) or amplitude ratio. Default value is 0.

stop_silence

Specify max duration of silence at end that will be kept after trimming. Default is 0, which is equal to trimming all samples detected as silence.

stop_mode

Specify mode of detection of silence start in end of multi-channel audio. Can be any or all. Default is any. With any, any sample that is detected as non-silence will cause stopped trimming of silence. With all, only if all channels are detected as non-silence will cause stopped trimming of silence.

detection

Set how is silence detected. Can be rms or peak. Second is faster and works better with digital silence which is exactly 0. Default value is rms.

window

Set duration in number of seconds used to calculate size of window in number of samples for detecting silence. Default value is 0.02. Allowed range is from 0 to 10.

Examples

• The following example shows how this filter can be used to start a recording that does not contain the delay at the start which usually occurs between pressing the record button and the start of the performance: silenceremove=start_periods=1:start_duration=5:start_threshold=0.02
• Trim all silence encountered from beginning to end where there is more than 1 second of silence in audio: silenceremove=stop_periods=-1:stop_duration=1:stop_threshold=-90dB
• Trim all digital silence samples, using peak detection, from beginning to end where there is more than 0 samples of digital silence in audio and digital silence is detected in all channels at same positions in stream: silenceremove=window=0:detection=peak:stop_mode=all:start_mode=all:stop_periods=-1:stop_threshold=0

SOFAlizer uses head-related transfer functions (HRTFs) to create virtual loudspeakers around the user for binaural listening via headphones (audio formats up to 9 channels supported). The HRTFs are stored in SOFA files (see <*http://www.sofacoustics.org/*> for a database). SOFAlizer is developed at the Acoustics Research Institute (ARI) of the Austrian Academy of Sciences.

To enable compilation of this filter you need to configure FFmpeg with --enable-libmysofa.

The filter accepts the following options:

sofa

Set the SOFA file used for rendering.

gain

Set gain applied to audio. Value is in dB. Default is 0.

rotation

Set rotation of virtual loudspeakers in deg. Default is 0.

elevation

Set elevation of virtual speakers in deg. Default is 0.

radius

Set distance in meters between loudspeakers and the listener with near-field HRTFs. Default is 1.

type

Set processing type. Can be time or freq. time is processing audio in time domain which is slow. freq is processing audio in frequency domain which is fast. Default is freq.

speakers

Set custom positions of virtual loudspeakers. Syntax for this option is: <CH> <AZIM> <ELEV>[|<CH> <AZIM> <ELEV>|…]. Each virtual loudspeaker is described with short channel name following with azimuth and elevation in degrees. Each virtual loudspeaker description is separated by ’|’. For example to override front left and front right channel positions use: ’speakers=FL 45 15|FR 345 15’. Descriptions with unrecognised channel names are ignored.

lfegain

Set custom gain for LFE channels. Value is in dB. Default is 0.

framesize

Set custom frame size in number of samples. Default is 1024. Allowed range is from 1024 to 96000. Only used if option type is set to freq.

normalize

Should all IRs be normalized upon importing SOFA file. By default is enabled.

interpolate

Should nearest IRs be interpolated with neighbor IRs if exact position does not match. By default is disabled.

minphase

Minphase all IRs upon loading of SOFA file. By default is disabled.

anglestep

Set neighbor search angle step. Only used if option interpolate is enabled.

radstep

Set neighbor search radius step. Only used if option interpolate is enabled.

Examples

• Using ClubFritz6 sofa file: sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=1
• Using ClubFritz12 sofa file and bigger radius with small rotation: sofalizer=sofa=/path/to/ClubFritz12.sofa:type=freq:radius=2:rotation=5
• Similar as above but with custom speaker positions for front left, front right, back left and back right and also with custom gain: “sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=2:speakers=FL 45|FR 315|BL 135|BR 225:gain=28”

Speech Normalizer.

This filter expands or compresses each half-cycle of audio samples (local set of samples all above or all below zero and between two nearest zero crossings) depending on threshold value, so audio reaches target peak value under conditions controlled by below options.

The filter accepts the following options:

peak, p

Set the expansion target peak value. This specifies the highest allowed absolute amplitude level for the normalized audio input. Default value is 0.95. Allowed range is from 0.0 to 1.0.

expansion, e

Set the maximum expansion factor. Allowed range is from 1.0 to 50.0. Default value is 2.0. This option controls maximum local half-cycle of samples expansion. The maximum expansion would be such that local peak value reaches target peak value but never to surpass it and that ratio between new and previous peak value does not surpass this option value.

compression, c

Set the maximum compression factor. Allowed range is from 1.0 to 50.0. Default value is 2.0. This option controls maximum local half-cycle of samples compression. This option is used only if threshold option is set to value greater than 0.0, then in such cases when local peak is lower or same as value set by threshold all samples belonging to that peak’s half-cycle will be compressed by current compression factor.

threshold, t

Set the threshold value. Default value is 0.0. Allowed range is from 0.0 to 1.0. This option specifies which half-cycles of samples will be compressed and which will be expanded. Any half-cycle samples with their local peak value below or same as this option value will be compressed by current compression factor, otherwise, if greater than threshold value they will be expanded with expansion factor so that it could reach peak target value but never surpass it.

raise, r

Set the expansion raising amount per each half-cycle of samples. Default value is 0.001. Allowed range is from 0.0 to 1.0. This controls how fast expansion factor is raised per each new half-cycle until it reaches expansion value. Setting this options too high may lead to distortions.

fall, f

Set the compression raising amount per each half-cycle of samples. Default value is 0.001. Allowed range is from 0.0 to 1.0. This controls how fast compression factor is raised per each new half-cycle until it reaches compression value.

channels, h

Specify which channels to filter, by default all available channels are filtered.

invert, i

Enable inverted filtering, by default is disabled. This inverts interpretation of threshold option. When enabled any half-cycle of samples with their local peak value below or same as threshold option will be expanded otherwise it will be compressed.

link, l

Link channels when calculating gain applied to each filtered channel sample, by default is disabled. When disabled each filtered channel gain calculation is independent, otherwise when this option is enabled the minimum of all possible gains for each filtered channel is used.

Commands

This filter supports the all above options as commands.

This filter has some handy utilities to manage stereo signals, for converting M/S stereo recordings to L/R signal while having control over the parameters or spreading the stereo image of master track.

The filter accepts the following options:

level_in

Set input level before filtering for both channels. Defaults is 1. Allowed range is from 0.015625 to 64.

level_out

Set output level after filtering for both channels. Defaults is 1. Allowed range is from 0.015625 to 64.

balance_in

Set input balance between both channels. Default is 0. Allowed range is from -1 to 1.

balance_out

Set output balance between both channels. Default is 0. Allowed range is from -1 to 1.

softclip

Enable softclipping. Results in analog distortion instead of harsh digital 0dB clipping. Disabled by default.

mutel

Mute the left channel. Disabled by default.

muter

Mute the right channel. Disabled by default.

phasel

Change the phase of the left channel. Disabled by default.

phaser

Change the phase of the right channel. Disabled by default.

mode

Set stereo mode. Available values are:

lr>lr

Left/Right to Left/Right, this is default.

lr>ms

Left/Right to Mid/Side.

ms>lr

Mid/Side to Left/Right.

lr>ll

Left/Right to Left/Left.

lr>rr

Left/Right to Right/Right.

lr>l+r

Left/Right to Left + Right.

lr>rl

Left/Right to Right/Left.

ms>ll

Mid/Side to Left/Left.

ms>rr

Mid/Side to Right/Right.

ms>rl

Mid/Side to Right/Left.

lr>l-r

Left/Right to Left - Right.

slev

Set level of side signal. Default is 1. Allowed range is from 0.015625 to 64.

sbal

Set balance of side signal. Default is 0. Allowed range is from -1 to 1.

mlev

Set level of the middle signal. Default is 1. Allowed range is from 0.015625 to 64.

mpan

Set middle signal pan. Default is 0. Allowed range is from -1 to 1.

base

Set stereo base between mono and inversed channels. Default is 0. Allowed range is from -1 to 1.

delay

Set delay in milliseconds how much to delay left from right channel and vice versa. Default is 0. Allowed range is from -20 to 20.

sclevel

Set S/C level. Default is 1. Allowed range is from 1 to 100.

phase

Set the stereo phase in degrees. Default is 0. Allowed range is from 0 to 360.

bmode_in, bmode_out

Set balance mode for balance_in/balance_out option. Can be one of the following:

balance

Classic balance mode. Attenuate one channel at time. Gain is raised up to 1.

amplitude

Similar as classic mode above but gain is raised up to 2.

power

Equal power distribution, from -6dB to +6dB range.

Commands

This filter supports the all above options as commands.

Examples

• Apply karaoke like effect: stereotools=mlev=0.015625
• Convert M/S signal to L/R: “stereotools=mode=ms>lr”

This filter enhance the stereo effect by suppressing signal common to both channels and by delaying the signal of left into right and vice versa, thereby widening the stereo effect.

The filter accepts the following options:

delay

Time in milliseconds of the delay of left signal into right and vice versa. Default is 20 milliseconds.

feedback

Amount of gain in delayed signal into right and vice versa. Gives a delay effect of left signal in right output and vice versa which gives widening effect. Default is 0.3.

crossfeed

Cross feed of left into right with inverted phase. This helps in suppressing the mono. If the value is 1 it will cancel all the signal common to both channels. Default is 0.3.

drymix

Set level of input signal of original channel. Default is 0.8.

Commands

This filter supports the all above options except delay as commands.

Apply 18 band equalizer.

The filter accepts the following options:

1b

Set 65Hz band gain.

2b

Set 92Hz band gain.

3b

Set 131Hz band gain.

4b

Set 185Hz band gain.

5b

Set 262Hz band gain.

6b

Set 370Hz band gain.

7b

Set 523Hz band gain.

8b

Set 740Hz band gain.

9b

Set 1047Hz band gain.

10b

Set 1480Hz band gain.

11b

Set 2093Hz band gain.

12b

Set 2960Hz band gain.

13b

Set 4186Hz band gain.

14b

Set 5920Hz band gain.

15b

Set 8372Hz band gain.

16b

Set 11840Hz band gain.

17b

Set 16744Hz band gain.

18b

Set 20000Hz band gain.

Apply audio surround upmix filter.

This filter allows to produce multichannel output from audio stream.

The filter accepts the following options:

chl_out

Set output channel layout. By default, this is 5.1. See the Channel Layout section in the ffmpeg-utils (1) manual for the required syntax.

chl_in

Set input channel layout. By default, this is stereo. See the Channel Layout section in the ffmpeg-utils (1) manual for the required syntax.

level_in

Set input volume level. By default, this is 1.

level_out

Set output volume level. By default, this is 1.

lfe

Enable LFE channel output if output channel layout has it. By default, this is enabled.

lfe_low

Set LFE low cut off frequency. By default, this is 128 Hz.

lfe_high

Set LFE high cut off frequency. By default, this is 256 Hz.

lfe_mode

Set LFE mode, can be add or sub. Default is add. In add mode, LFE channel is created from input audio and added to output. In sub mode, LFE channel is created from input audio and added to output but also all non-LFE output channels are subtracted with output LFE channel.

angle

Set angle of stereo surround transform, Allowed range is from 0 to 360. Default is 90.

fc_in

Set front center input volume. By default, this is 1.

fc_out

Set front center output volume. By default, this is 1.

fl_in

Set front left input volume. By default, this is 1.

fl_out

Set front left output volume. By default, this is 1.

fr_in

Set front right input volume. By default, this is 1.

fr_out

Set front right output volume. By default, this is 1.

sl_in

Set side left input volume. By default, this is 1.

sl_out

Set side left output volume. By default, this is 1.

sr_in

Set side right input volume. By default, this is 1.

sr_out

Set side right output volume. By default, this is 1.

bl_in

Set back left input volume. By default, this is 1.

bl_out

Set back left output volume. By default, this is 1.

br_in

Set back right input volume. By default, this is 1.

br_out

Set back right output volume. By default, this is 1.

bc_in

Set back center input volume. By default, this is 1.

bc_out

Set back center output volume. By default, this is 1.

lfe_in

Set LFE input volume. By default, this is 1.

lfe_out

Set LFE output volume. By default, this is 1.

allx

Set spread usage of stereo image across X axis for all channels.

ally

Set spread usage of stereo image across Y axis for all channels.

fcx, flx, frx, blx, brx, slx, srx, bcx

Set spread usage of stereo image across X axis for each channel.

fcy, fly, fry, bly, bry, sly, sry, bcy

Set spread usage of stereo image across Y axis for each channel.

win_size

Set window size. Allowed range is from 1024 to 65536. Default size is 4096.

win_func

Set window function. It accepts the following values:

rect

bartlett

hann, hanning

hamming

blackman

welch

flattop

bharris

bnuttall

bhann

sine

nuttall

lanczos

gauss

tukey

dolph

cauchy

parzen

poisson

bohman

Default is hann.

overlap

Set window overlap. If set to 1, the recommended overlap for selected window function will be picked. Default is 0.5.

Boost or cut treble (upper) frequencies of the audio using a two-pole shelving filter with a response similar to that of a standard hi-fi’s tone-controls. This is also known as shelving equalisation (EQ).

The filter accepts the following options:

gain, g

Give the gain at whichever is the lower of ~22 kHz and the Nyquist frequency. Its useful range is about -20 (for a large cut) to +20 (for a large boost). Beware of clipping when using a positive gain.

frequency, f

Set the filter’s central frequency and so can be used to extend or reduce the frequency range to be boosted or cut. The default value is 3000 Hz.

width_type, t

Set method to specify band-width of filter.

1. Hz
2. Q-Factor
3. octave
4. slope
5. kHz

width, w

Determine how steep is the filter’s shelf transition.

poles, p

Set number of poles. Default is 2.

mix, m

How much to use filtered signal in output. Default is 1. Range is between 0 and 1.

channels, c

Specify which channels to filter, by default all available are filtered.

normalize, n

Normalize biquad coefficients, by default is disabled. Enabling it will normalize magnitude response at DC to 0dB.

transform, a

Set transform type of IIR filter.

precision, r

Set precison of filtering.

auto

Pick automatic sample format depending on surround filters.

s16

Always use signed 16-bit.

s32

Always use signed 32-bit.

f32

Always use float 32-bit.

f64

Always use float 64-bit.

nil

Commands

This filter supports the following commands:

frequency, f

Change treble frequency. Syntax for the command is : “frequency”

width_type, t

Change treble width_type. Syntax for the command is : “width_type”

width, w

Change treble width. Syntax for the command is : “width”

gain, g

Change treble gain. Syntax for the command is : “gain”

mix, m

Change treble mix. Syntax for the command is : “mix”

Sinusoidal amplitude modulation.

The filter accepts the following options:

1. Modulation frequency in Hertz. Modulation frequencies in the subharmonic range (20 Hz or lower) will result in a tremolo effect. This filter may also be used as a ring modulator by specifying a modulation frequency higher than 20 Hz. Range is 0.1 - 20000.0. Default value is 5.0 Hz.
2. Depth of modulation as a percentage. Range is 0.0 - 1.0. Default value is 0.5.

Sinusoidal phase modulation.

The filter accepts the following options:

1. Modulation frequency in Hertz. Range is 0.1 - 20000.0. Default value is 5.0 Hz.
2. Depth of modulation as a percentage. Range is 0.0 - 1.0. Default value is 0.5.

Adjust the input audio volume.

It accepts the following parameters:

volume

Set audio volume expression. Output values are clipped to the maximum value. The output audio volume is given by the relation: <output_volume> = <volume> * <input_volume> The default value for volume is 1.0.

precision

This parameter represents the mathematical precision. It determines which input sample formats will be allowed, which affects the precision of the volume scaling.

fixed

8-bit fixed-point; this limits input sample format to U8, S16, and S32.

float

32-bit floating-point; this limits input sample format to FLT. (default)

double

64-bit floating-point; this limits input sample format to DBL.

replaygain

Choose the behaviour on encountering ReplayGain side data in input frames.

drop

Remove ReplayGain side data, ignoring its contents (the default).

ignore

Ignore ReplayGain side data, but leave it in the frame.

track

Prefer the track gain, if present.

album

Prefer the album gain, if present.

replaygain_preamp

Pre-amplification gain in dB to apply to the selected replaygain gain. Default value for replaygain_preamp is 0.0.

replaygain_noclip

Prevent clipping by limiting the gain applied. Default value for replaygain_noclip is 1.

eval

Set when the volume expression is evaluated. It accepts the following values:

once

only evaluate expression once during the filter initialization, or when the volume command is sent

frame

evaluate expression for each incoming frame

Default value is once.

The volume expression can contain the following parameters.

1. frame number (starting at zero)
2. number of channels
3. number of samples consumed by the filter
4. number of samples in the current frame
5. original frame position in the file
6. frame PTS
7. sample rate
8. PTS at start of stream
9. time at start of stream
10. frame time
11. timestamp timebase
12. last set volume value

Note that when eval is set to once only the sample_rate and tb variables are available, all other variables will evaluate to NAN.

Commands

This filter supports the following commands:

volume

Modify the volume expression. The command accepts the same syntax of the corresponding option. If the specified expression is not valid, it is kept at its current value.

Examples

• Halve the input audio volume: volume=volume=0.5 volume=volume=1/2 volume=volume=-6.0206dB In all the above example the named key for volume can be omitted, for example like in: volume=0.5
• Increase input audio power by 6 decibels using fixed-point precision: volume=volume=6dB:precision=fixed
• Fade volume after time 10 with an annihilation period of 5 seconds: volume=if(lt(t,10),1,max(1-(t-10)/5,0)):eval=frame

Detect the volume of the input video.

The filter has no parameters. The input is not modified. Statistics about the volume will be printed in the log when the input stream end is reached.

In particular it will show the mean volume (root mean square), maximum volume (on a per-sample basis), and the beginning of a histogram of the registered volume values (from the maximum value to a cumulated 1/1000 of the samples).

All volumes are in decibels relative to the maximum PCM value.

Examples

Here is an excerpt of the output:

[Parsed_volumedetect_0 0xa23120] mean_volume: -27 dB [Parsed_volumedetect_0 0xa23120] max_volume: -4 dB [Parsed_volumedetect_0 0xa23120] histogram_4db: 6 [Parsed_volumedetect_0 0xa23120] histogram_5db: 62 [Parsed_volumedetect_0 0xa23120] histogram_6db: 286 [Parsed_volumedetect_0 0xa23120] histogram_7db: 1042 [Parsed_volumedetect_0 0xa23120] histogram_8db: 2551 [Parsed_volumedetect_0 0xa23120] histogram_9db: 4609 [Parsed_volumedetect_0 0xa23120] histogram_10db: 8409

It means that:

• The mean square energy is approximately -27 dB, or 10^-2.7.
• The largest sample is at -4 dB, or more precisely between -4 dB and -5 dB.
• There are 6 samples at -4 dB, 62 at -5 dB, 286 at -6 dB, etc.

In other words, raising the volume by +4 dB does not cause any clipping, raising it by +5 dB causes clipping for 6 samples, etc.

Buffer audio frames, and make them available to the filter chain.

This source is mainly intended for a programmatic use, in particular through the interface defined in libavfilter/buffersrc.h.

It accepts the following parameters:

time_base

The timebase which will be used for timestamps of submitted frames. It must be either a floating-point number or in numerator///denominator form.

sample_rate

The sample rate of the incoming audio buffers.

sample_fmt

The sample format of the incoming audio buffers. Either a sample format name or its corresponding integer representation from the enum AVSampleFormat in libavutil/samplefmt.h

channel_layout

The channel layout of the incoming audio buffers. Either a channel layout name from channel_layout_map in libavutil/channel_layout.c or its corresponding integer representation from the AV_CH_LAYOUT_* macros in libavutil/channel_layout.h

channels

The number of channels of the incoming audio buffers. If both channels and channel_layout are specified, then they must be consistent.

Examples

abuffer=sample_rate=44100:sample_fmt=s16p:channel_layout=stereo

will instruct the source to accept planar 16bit signed stereo at 44100Hz. Since the sample format with name s16p corresponds to the number 6 and the stereo channel layout corresponds to the value 0x3, this is equivalent to:

abuffer=sample_rate=44100:sample_fmt=6:channel_layout=0x3

Generate an audio signal specified by an expression.

This source accepts in input one or more expressions (one for each channel), which are evaluated and used to generate a corresponding audio signal.

This source accepts the following options:

exprs

Set the ’|’-separated expressions list for each separate channel. In case the channel_layout option is not specified, the selected channel layout depends on the number of provided expressions. Otherwise the last specified expression is applied to the remaining output channels.

channel_layout, c

Set the channel layout. The number of channels in the specified layout must be equal to the number of specified expressions.

duration, d

Set the minimum duration of the sourced audio. See the Time duration section in the ffmpeg-utils (1) manual for the accepted syntax. Note that the resulting duration may be greater than the specified duration, as the generated audio is always cut at the end of a complete frame. If not specified, or the expressed duration is negative, the audio is supposed to be generated forever.

nb_samples, n

Set the number of samples per channel per each output frame, default to 1024.

sample_rate, s

Specify the sample rate, default to 44100.

Each expression in exprs can contain the following constants:

1. number of the evaluated sample, starting from 0
2. time of the evaluated sample expressed in seconds, starting from 0
3. sample rate

Examples

• Generate silence: aevalsrc=0
• Generate a sin signal with frequency of 440 Hz, set sample rate to 8000 Hz: aevalsrc=“sin(440*2*PI*t):s=8000”
• Generate a two channels signal, specify the channel layout (Front Center + Back Center) explicitly: aevalsrc=“sin(420*2*PI*t)|cos(430*2*PI*t):c=FC|BC”
• Generate white noise: aevalsrc=“-2+random(0)”
• Generate an amplitude modulated signal: aevalsrc=“sin(10*2*PI*t)*sin(880*2*PI*t)”
• Generate 2.5 Hz binaural beats on a 360 Hz carrier: aevalsrc=“0.1*sin(2*PI*(360-2.5/2)*t) | 0.1*sin(2*PI*(360+2.5/2)*t)”

Generate a FIR coefficients using frequency sampling method.

The resulting stream can be used with afir filter for filtering the audio signal.

The filter accepts the following options:

taps, t

Set number of filter coefficents in output audio stream. Default value is 1025.

frequency, f

Set frequency points from where magnitude and phase are set. This must be in non decreasing order, and first element must be 0, while last element must be 1. Elements are separated by white spaces.

magnitude, m

Set magnitude value for every frequency point set by frequency. Number of values must be same as number of frequency points. Values are separated by white spaces.

phase, p

Set phase value for every frequency point set by frequency. Number of values must be same as number of frequency points. Values are separated by white spaces.

sample_rate, r

Set sample rate, default is 44100.

nb_samples, n

Set number of samples per each frame. Default is 1024.

win_func, w

Set window function. Default is blackman.

The null audio source, return unprocessed audio frames. It is mainly useful as a template and to be employed in analysis / debugging tools, or as the source for filters which ignore the input data (for example the sox synth filter).

This source accepts the following options:

channel_layout, cl

Specifies the channel layout, and can be either an integer or a string representing a channel layout. The default value of channel_layout is stereo. Check the channel_layout_map definition in libavutil/channel_layout.c for the mapping between strings and channel layout values.

sample_rate, r

Specifies the sample rate, and defaults to 44100.

nb_samples, n

Set the number of samples per requested frames.

duration, d

Set the duration of the sourced audio. See the Time duration section in the ffmpeg-utils (1) manual for the accepted syntax. If not specified, or the expressed duration is negative, the audio is supposed to be generated forever.

Examples

• Set the sample rate to 48000 Hz and the channel layout to AV_CH_LAYOUT_MONO. anullsrc=r=48000:cl=4
• Do the same operation with a more obvious syntax: anullsrc=r=48000:cl=mono

All the parameters need to be explicitly defined.

Synthesize a voice utterance using the libflite library.

To enable compilation of this filter you need to configure FFmpeg with --enable-libflite.

Note that versions of the flite library prior to 2.0 are not thread-safe.

The filter accepts the following options:

list_voices

If set to 1, list the names of the available voices and exit immediately. Default value is 0.

nb_samples, n

Set the maximum number of samples per frame. Default value is 512.

textfile

Set the filename containing the text to speak.

text

Set the text to speak.

voice, v

Set the voice to use for the speech synthesis. Default value is kal. See also the list_voices option.

Examples

• Read from file speech.txt, and synthesize the text using the standard flite voice: flite=textfile=speech.txt
• Read the specified text selecting the slt voice: flite=text=So fare thee well, poor devil of a Sub-Sub, whose commentator I am:voice=slt
• Input text to ffmpeg: ffmpeg -f lavfi -i flite=text=So fare thee well, poor devil of a Sub-Sub, whose commentator I am:voice=slt
• Make ffplay speak the specified text, using flite and the lavfi device: ffplay -f lavfi flite=text=No more be grieved for which that thou hast done.

For more information about libflite, check: <*http://www.festvox.org/flite/*>

Generate a noise audio signal.

The filter accepts the following options:

sample_rate, r

Specify the sample rate. Default value is 48000 Hz.

amplitude, a

Specify the amplitude (0.0 - 1.0) of the generated audio stream. Default value is 1.0.

duration, d

Specify the duration of the generated audio stream. Not specifying this option results in noise with an infinite length.

color, colour, c

Specify the color of noise. Available noise colors are white, pink, brown, blue, violet and velvet. Default color is white.

seed, s

Specify a value used to seed the PRNG.

nb_samples, n

Set the number of samples per each output frame, default is 1024.

Examples

• Generate 60 seconds of pink noise, with a 44.1 kHz sampling rate and an amplitude of 0.5: anoisesrc=d=60:c=pink:r=44100:a=0.5

Generate odd-tap Hilbert transform FIR coefficients.

The resulting stream can be used with afir filter for phase-shifting the signal by 90 degrees.

This is used in many matrix coding schemes and for analytic signal generation. The process is often written as a multiplication by i (or j), the imaginary unit.

The filter accepts the following options:

sample_rate, s

Set sample rate, default is 44100.

taps, t

Set length of FIR filter, default is 22051.

nb_samples, n

Set number of samples per each frame.

win_func, w

Set window function to be used when generating FIR coefficients.

Generate a sinc kaiser-windowed low-pass, high-pass, band-pass, or band-reject FIR coefficients.

The resulting stream can be used with afir filter for filtering the audio signal.

The filter accepts the following options:

sample_rate, r

Set sample rate, default is 44100.

nb_samples, n

Set number of samples per each frame. Default is 1024.

hp

Set high-pass frequency. Default is 0.

lp

Set low-pass frequency. Default is 0. If high-pass frequency is lower than low-pass frequency and low-pass frequency is higher than 0 then filter will create band-pass filter coefficients, otherwise band-reject filter coefficients.

phase

Set filter phase response. Default is 50. Allowed range is from 0 to 100.

beta

Set Kaiser window beta.

att

Set stop-band attenuation. Default is 120dB, allowed range is from 40 to 180 dB.

round

Enable rounding, by default is disabled.

hptaps

Set number of taps for high-pass filter.

lptaps

Set number of taps for low-pass filter.

Generate an audio signal made of a sine wave with amplitude 1/8.

The audio signal is bit-exact.

The filter accepts the following options:

frequency, f

Set the carrier frequency. Default is 440 Hz.

beep_factor, b

Enable a periodic beep every second with frequency beep_factor times the carrier frequency. Default is 0, meaning the beep is disabled.

sample_rate, r

Specify the sample rate, default is 44100.

duration, d

Specify the duration of the generated audio stream.

samples_per_frame

Set the number of samples per output frame. The expression can contain the following constants:

1. The (sequential) number of the output audio frame, starting from 0.
2. The PTS (Presentation TimeStamp) of the output audio frame, expressed in TB units.
3. The PTS of the output audio frame, expressed in seconds.
4. The timebase of the output audio frames.

Default is 1024.

Examples

• Generate a simple 440 Hz sine wave: sine
• Generate a 220 Hz sine wave with a 880 Hz beep each second, for 5 seconds: sine=220:4:d=5 sine=f=220:b=4:d=5 sine=frequency=220:beep_factor=4:duration=5
• Generate a 1 kHz sine wave following 1602,1601,1602,1601,1602 NTSC pattern: sine=1000:samples_per_frame=st(0,mod(n,5)); 1602-not(not(eq(ld(0),1)+eq(ld(0),3)))

Buffer audio frames, and make them available to the end of filter chain.

This sink is mainly intended for programmatic use, in particular through the interface defined in libavfilter/buffersink.h or the options system.

It accepts a pointer to an AVABufferSinkContext structure, which defines the incoming buffers’ formats, to be passed as the opaque parameter to avfilter_init_filter for initialization.

Null audio sink; do absolutely nothing with the input audio. It is mainly useful as a template and for use in analysis / debugging tools.

Mark a region of interest in a video frame.

The frame data is passed through unchanged, but metadata is attached to the frame indicating regions of interest which can affect the behaviour of later encoding. Multiple regions can be marked by applying the filter multiple times.

1. Region distance in pixels from the left edge of the frame.
2. Region distance in pixels from the top edge of the frame.
3. Region width in pixels.
4. Region height in pixels. The parameters x, y, w and h are expressions, and may contain the following variables:

   iw

   Width of the input frame.

   ih

   Height of the input frame.

5. Quantisation offset to apply within the region. This must be a real value in the range -1 to +1. A value of zero indicates no quality change. A negative value asks for better quality (less quantisation), while a positive value asks for worse quality (greater quantisation). The range is calibrated so that the extreme values indicate the largest possible offset - if the rest of the frame is encoded with the worst possible quality, an offset of -1 indicates that this region should be encoded with the best possible quality anyway. Intermediate values are then interpolated in some codec-dependent way. For example, in 10-bit H.264 the quantisation parameter varies between -12 and 51. A typical qoffset value of -1/10 therefore indicates that this region should be encoded with a QP around one-tenth of the full range better than the rest of the frame. So, if most of the frame were to be encoded with a QP of around 30, this region would get a QP of around 24 (an offset of approximately -1/10 * (51 - -12) = -6.3). An extreme value of -1 would indicate that this region should be encoded with the best possible quality regardless of the treatment of the rest of the frame - that is, should be encoded at a QP of -12.
6. If set to true, remove any existing regions of interest marked on the frame before adding the new one.

Examples

• Mark the centre quarter of the frame as interesting. addroi=iw/4:ih/4:iw/2:ih/2:-1/10
• Mark the 100-pixel-wide region on the left edge of the frame as very uninteresting (to be encoded at much lower quality than the rest of the frame). addroi=0:0:100:ih:+1/5

Extract the alpha component from the input as a grayscale video. This is especially useful with the alphamerge filter.

Add or replace the alpha component of the primary input with the grayscale value of a second input. This is intended for use with alphaextract to allow the transmission or storage of frame sequences that have alpha in a format that doesn’t support an alpha channel.

For example, to reconstruct full frames from a normal YUV-encoded video and a separate video created with alphaextract, you might use:

movie=in_alpha.mkv [alpha]; [in][alpha] alphamerge [out]

Amplify differences between current pixel and pixels of adjacent frames in same pixel location.

This filter accepts the following options:

radius

Set frame radius. Default is 2. Allowed range is from 1 to 63. For example radius of 3 will instruct filter to calculate average of 7 frames.

factor

Set factor to amplify difference. Default is 2. Allowed range is from 0 to 65535.

threshold

Set threshold for difference amplification. Any difference greater or equal to this value will not alter source pixel. Default is 10. Allowed range is from 0 to 65535.

tolerance

Set tolerance for difference amplification. Any difference lower to this value will not alter source pixel. Default is 0. Allowed range is from 0 to 65535.

low

Set lower limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535. This option controls maximum possible value that will decrease source pixel value.

high

Set high limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535. This option controls maximum possible value that will increase source pixel value.

planes

Set which planes to filter. Default is all. Allowed range is from 0 to 15.

Commands

This filter supports the following commands that corresponds to option of same name:

factor

threshold

tolerance

low

high

planes

Same as the subtitles filter, except that it doesn’t require libavcodec and libavformat to work. On the other hand, it is limited to ASS (Advanced Substation Alpha) subtitles files.

This filter accepts the following option in addition to the common options from the subtitles filter:

shaping

Set the shaping engine Available values are:

auto

The default libass shaping engine, which is the best available.

simple

Fast, font-agnostic shaper that can do only substitutions

complex

Slower shaper using OpenType for substitutions and positioning

The default is auto.

Apply an Adaptive Temporal Averaging Denoiser to the video input.

The filter accepts the following options:

0a

Set threshold A for 1st plane. Default is 0.02. Valid range is 0 to 0.3.

0b

Set threshold B for 1st plane. Default is 0.04. Valid range is 0 to 5.

1a

Set threshold A for 2nd plane. Default is 0.02. Valid range is 0 to 0.3.

1b

Set threshold B for 2nd plane. Default is 0.04. Valid range is 0 to 5.

2a

Set threshold A for 3rd plane. Default is 0.02. Valid range is 0 to 0.3.

2b

Set threshold B for 3rd plane. Default is 0.04. Valid range is 0 to 5. Threshold A is designed to react on abrupt changes in the input signal and threshold B is designed to react on continuous changes in the input signal.

(no term)

Set number of frames filter will use for averaging. Default is 9. Must be odd number in range [5, 129].

(no term)

Set what planes of frame filter will use for averaging. Default is all.

(no term)

Set what variant of algorithm filter will use for averaging. Default is p parallel. Alternatively can be set to s serial. Parallel can be faster then serial, while other way around is never true. Parallel will abort early on first change being greater then thresholds, while serial will continue processing other side of frames if they are equal or below thresholds.

0s

1s

2s

Set sigma for 1st plane, 2nd plane or 3rd plane. Default is 32767. Valid range is from 0 to 32767. This options controls weight for each pixel in radius defined by size. Default value means every pixel have same weight. Setting this option to 0 effectively disables filtering.

Commands

This filter supports same commands as options except option s. The command accepts the same syntax of the corresponding option.

Apply average blur filter.

The filter accepts the following options:

sizeX

Set horizontal radius size.

planes

Set which planes to filter. By default all planes are filtered.

sizeY

Set vertical radius size, if zero it will be same as sizeX. Default is 0.

Commands

This filter supports same commands as options. The command accepts the same syntax of the corresponding option.

If the specified expression is not valid, it is kept at its current value.

Compute the bounding box for the non-black pixels in the input frame luminance plane.

This filter computes the bounding box containing all the pixels with a luminance value greater than the minimum allowed value. The parameters describing the bounding box are printed on the filter log.

The filter accepts the following option:

min_val

Set the minimal luminance value. Default is 16.

Commands

This filter supports the all above options as commands.

Apply bilateral filter, spatial smoothing while preserving edges.

The filter accepts the following options:

sigmaS

Set sigma of gaussian function to calculate spatial weight. Allowed range is 0 to 512. Default is 0.1.

sigmaR

Set sigma of gaussian function to calculate range weight. Allowed range is 0 to 1. Default is 0.1.

planes

Set planes to filter. Default is first only.

Commands

This filter supports the all above options as commands.

Show and measure bit plane noise.

The filter accepts the following options:

bitplane

Set which plane to analyze. Default is 1.

filter

Filter out noisy pixels from bitplane set above. Default is disabled.

Detect video intervals that are (almost) completely black. Can be useful to detect chapter transitions, commercials, or invalid recordings.

The filter outputs its detection analysis to both the log as well as frame metadata. If a black segment of at least the specified minimum duration is found, a line with the start and end timestamps as well as duration is printed to the log with level info. In addition, a log line with level debug is printed per frame showing the black amount detected for that frame.

The filter also attaches metadata to the first frame of a black segment with key lavfi.black_start and to the first frame after the black segment ends with key lavfi.black_end. The value is the frame’s timestamp. This metadata is added regardless of the minimum duration specified.

The filter accepts the following options:

black_min_duration, d

Set the minimum detected black duration expressed in seconds. It must be a non-negative floating point number. Default value is 2.0.

picture_black_ratio_th, pic_th

Set the threshold for considering a picture black. Express the minimum value for the ratio: <nb_black_pixels> / <nb_pixels> for which a picture is considered black. Default value is 0.98.

pixel_black_th, pix_th

Set the threshold for considering a pixel black. The threshold expresses the maximum pixel luminance value for which a pixel is considered black. The provided value is scaled according to the following equation: <absolute_threshold> = <luminance_minimum_value> + <pixel_black_th> * <luminance_range_size> luminance_range_size and luminance_minimum_value depend on the input video format, the range is [0-255] for YUV full-range formats and [16-235] for YUV non full-range formats. Default value is 0.10.

The following example sets the maximum pixel threshold to the minimum value, and detects only black intervals of 2 or more seconds:

blackdetect=d=2:pix_th=0.00

Detect frames that are (almost) completely black. Can be useful to detect chapter transitions or commercials. Output lines consist of the frame number of the detected frame, the percentage of blackness, the position in the file if known or -1 and the timestamp in seconds.

In order to display the output lines, you need to set the loglevel at least to the AV_LOG_INFO value.

This filter exports frame metadata lavfi.blackframe.pblack. The value represents the percentage of pixels in the picture that are below the threshold value.

It accepts the following parameters:

amount

The percentage of the pixels that have to be below the threshold; it defaults to 98.

threshold, thresh

The threshold below which a pixel value is considered black; it defaults to 32.

Blend two video frames into each other.

The blend filter takes two input streams and outputs one stream, the first input is the top layer and second input is bottom layer. By default, the output terminates when the longest input terminates.

The tblend (time blend) filter takes two consecutive frames from one single stream, and outputs the result obtained by blending the new frame on top of the old frame.

A description of the accepted options follows.

c0_mode

c1_mode

c2_mode

c3_mode

all_mode

Set blend mode for specific pixel component or all pixel components in case of all_mode. Default value is normal. Available values for component modes are:

addition

grainmerge

and

average

burn

darken

difference

grainextract

divide

dodge

freeze

exclusion

extremity

glow

hardlight

hardmix

heat

lighten

linearlight

multiply

multiply128

negation

normal

or

overlay

phoenix

pinlight

reflect

screen

softlight

subtract

vividlight

xor

nil

c0_opacity

c1_opacity

c2_opacity

c3_opacity

all_opacity

Set blend opacity for specific pixel component or all pixel components in case of all_opacity. Only used in combination with pixel component blend modes.

c0_expr

c1_expr

c2_expr

c3_expr

all_expr

Set blend expression for specific pixel component or all pixel components in case of all_expr. Note that related mode options will be ignored if those are set. The expressions can use the following variables:

1. The sequential number of the filtered frame, starting from 0.

the coordinates of the current sample

the width and height of currently filtered plane

SW

SH

Width and height scale for the plane being filtered. It is the ratio between the dimensions of the current plane to the luma plane, e.g. for a yuv420p frame, the values are 1,1 for the luma plane and 0.5,0.5 for the chroma planes.

1. Time of the current frame, expressed in seconds.
2. Value of pixel component at current location for first video frame (top layer).
3. Value of pixel component at current location for second video frame (bottom layer).

nil

The blend filter also supports the framesync options.

Examples

• Apply transition from bottom layer to top layer in first 10 seconds: blend=all_expr=A*(if(gte(T,10),1,T/10))+B*(1-(if(gte(T,10),1,T/10)))
• Apply linear horizontal transition from top layer to bottom layer: blend=all_expr=A*(X/W)+B*(1-X/W)
• Apply 1x1 checkerboard effect: blend=all_expr=if(eq(mod(X,2),mod(Y,2)),A,B)
• Apply uncover left effect: blend=all_expr=if(gte(N*SW+X,W),A,B)
• Apply uncover down effect: blend=all_expr=if(gte(Y-N*SH,0),A,B)
• Apply uncover up-left effect: blend=all_expr=if(gte(T*SH*40+Y,H)*gte((T*40*SW+X)*W/H,W),A,B)
• Split diagonally video and shows top and bottom layer on each side: blend=all_expr=if(gt(X,Y*(W/H)),A,B)
• Display differences between the current and the previous frame: tblend=all_mode=grainextract

Commands

This filter supports same commands as options.

Denoise frames using Block-Matching 3D algorithm.

The filter accepts the following options.

sigma

Set denoising strength. Default value is 1. Allowed range is from 0 to 999.9. The denoising algorithm is very sensitive to sigma, so adjust it according to the source.

block

Set local patch size. This sets dimensions in 2D.

bstep

Set sliding step for processing blocks. Default value is 4. Allowed range is from 1 to 64. Smaller values allows processing more reference blocks and is slower.

group

Set maximal number of similar blocks for 3rd dimension. Default value is 1. When set to 1, no block matching is done. Larger values allows more blocks in single group. Allowed range is from 1 to 256.

range

Set radius for search block matching. Default is 9. Allowed range is from 1 to INT32_MAX.

mstep

Set step between two search locations for block matching. Default is 1. Allowed range is from 1 to 64. Smaller is slower.

thmse

Set threshold of mean square error for block matching. Valid range is 0 to INT32_MAX.

hdthr

Set thresholding parameter for hard thresholding in 3D transformed domain. Larger values results in stronger hard-thresholding filtering in frequency domain.

estim

Set filtering estimation mode. Can be basic or final. Default is basic.

ref

If enabled, filter will use 2nd stream for block matching. Default is disabled for basic value of estim option, and always enabled if value of estim is final.

planes

Set planes to filter. Default is all available except alpha.

Examples

• Basic filtering with bm3d: bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic
• Same as above, but filtering only luma: bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic:planes=1
• Same as above, but with both estimation modes: split[a][b],[a]bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic[a],[b][a]bm3d=sigma=3:block=4:bstep=2:group=16:estim=final:ref=1
• Same as above, but prefilter with nlmeans filter instead: split[a][b],[a]nlmeans=s=3:r=7:p=3[a],[b][a]bm3d=sigma=3:block=4:bstep=2:group=16:estim=final:ref=1

Apply a boxblur algorithm to the input video.

It accepts the following parameters:

luma_radius, lr

luma_power, lp

chroma_radius, cr

chroma_power, cp

alpha_radius, ar

alpha_power, ap

A description of the accepted options follows.

luma_radius, lr

chroma_radius, cr

alpha_radius, ar

Set an expression for the box radius in pixels used for blurring the corresponding input plane. The radius value must be a non-negative number, and must not be greater than the value of the expression min(w,h)/2 for the luma and alpha planes, and of min(cw,ch)/2 for the chroma planes. Default value for luma_radius is 2. If not specified, chroma_radius and alpha_radius default to the corresponding value set for luma_radius. The expressions can contain the following constants:

The input width and height in pixels.

cw

ch

The input chroma image width and height in pixels.

hsub

vsub

The horizontal and vertical chroma subsample values. For example, for the pixel format yuv422p, hsub is 2 and vsub is 1.

nil

luma_power, lp

chroma_power, cp

alpha_power, ap

Specify how many times the boxblur filter is applied to the corresponding plane. Default value for luma_power is 2. If not specified, chroma_power and alpha_power default to the corresponding value set for luma_power. A value of 0 will disable the effect.

Examples

• Apply a boxblur filter with the luma, chroma, and alpha radii set to 2: boxblur=luma_radius=2:luma_power=1 boxblur=2:1
• Set the luma radius to 2, and alpha and chroma radius to 0: boxblur=2:1:cr=0:ar=0
• Set the luma and chroma radii to a fraction of the video dimension: boxblur=luma_radius=min(h\,w)/10:luma_power=1:chroma_radius=min(cw\,ch)/10:chroma_power=1

Deinterlace the input video (bwdif stands for Bob Weaver Deinterlacing Filter).

Motion adaptive deinterlacing based on yadif with the use of w3fdif and cubic interpolation algorithms. It accepts the following parameters:

mode

The interlacing mode to adopt. It accepts one of the following values:

0, send_frame

Output one frame for each frame.

1, send_field

Output one frame for each field.

The default value is send_field.

parity

The picture field parity assumed for the input interlaced video. It accepts one of the following values:

0, tff

Assume the top field is first.

1, bff

Assume the bottom field is first.

-1, auto

Enable automatic detection of field parity.

The default value is auto. If the interlacing is unknown or the decoder does not export this information, top field first will be assumed.

deint

Specify which frames to deinterlace. Accepts one of the following values:

0, all

Deinterlace all frames.

1, interlaced

Only deinterlace frames marked as interlaced.

The default value is all.

Apply Contrast Adaptive Sharpen filter to video stream.

The filter accepts the following options:

strength

Set the sharpening strength. Default value is 0.

planes

Set planes to filter. Default value is to filter all planes except alpha plane.

Commands

This filter supports same commands as options.

Remove all color information for all colors except for certain one.

The filter accepts the following options:

color

The color which will not be replaced with neutral chroma.

similarity

Similarity percentage with the above color. 0.01 matches only the exact key color, while 1.0 matches everything.

blend

Blend percentage. 0.0 makes pixels either fully gray, or not gray at all. Higher values result in more preserved color.

yuv

Signals that the color passed is already in YUV instead of RGB. Literal colors like green or red don’t make sense with this enabled anymore. This can be used to pass exact YUV values as hexadecimal numbers.

Commands

This filter supports same commands as options. The command accepts the same syntax of the corresponding option.

If the specified expression is not valid, it is kept at its current value.

YUV colorspace color/chroma keying.

The filter accepts the following options:

color

The color which will be replaced with transparency.

similarity

Similarity percentage with the key color. 0.01 matches only the exact key color, while 1.0 matches everything.

blend

Blend percentage. 0.0 makes pixels either fully transparent, or not transparent at all. Higher values result in semi-transparent pixels, with a higher transparency the more similar the pixels color is to the key color.

yuv

Signals that the color passed is already in YUV instead of RGB. Literal colors like green or red don’t make sense with this enabled anymore. This can be used to pass exact YUV values as hexadecimal numbers.

Commands

This filter supports same commands as options. The command accepts the same syntax of the corresponding option.

If the specified expression is not valid, it is kept at its current value.

Examples

• Make every green pixel in the input image transparent: ffmpeg -i input.png -vf chromakey=green out.png
• Overlay a greenscreen-video on top of a static black background. ffmpeg -f lavfi -i color=c=black:s=1280x720 -i video.mp4 -shortest -filter_complex “[1:v]chromakey=0x70de77:0.1:0.2[ckout];[0:v][ckout]overlay[out]” -map “[out]” output.mkv

Reduce chrominance noise.

The filter accepts the following options:

thres

Set threshold for averaging chrominance values. Sum of absolute difference of Y, U and V pixel components of current pixel and neighbour pixels lower than this threshold will be used in averaging. Luma component is left unchanged and is copied to output. Default value is 30. Allowed range is from 1 to 200.

sizew

Set horizontal radius of rectangle used for averaging. Allowed range is from 1 to 100. Default value is 5.

sizeh

Set vertical radius of rectangle used for averaging. Allowed range is from 1 to 100. Default value is 5.

stepw

Set horizontal step when averaging. Default value is 1. Allowed range is from 1 to 50. Mostly useful to speed-up filtering.

steph

Set vertical step when averaging. Default value is 1. Allowed range is from 1 to 50. Mostly useful to speed-up filtering.

threy

Set Y threshold for averaging chrominance values. Set finer control for max allowed difference between Y components of current pixel and neigbour pixels. Default value is 200. Allowed range is from 1 to 200.

threu

Set U threshold for averaging chrominance values. Set finer control for max allowed difference between U components of current pixel and neigbour pixels. Default value is 200. Allowed range is from 1 to 200.

threv

Set V threshold for averaging chrominance values. Set finer control for max allowed difference between V components of current pixel and neigbour pixels. Default value is 200. Allowed range is from 1 to 200.

Commands

This filter supports same commands as options. The command accepts the same syntax of the corresponding option.

Shift chroma pixels horizontally and/or vertically.

The filter accepts the following options:

cbh

Set amount to shift chroma-blue horizontally.

cbv

Set amount to shift chroma-blue vertically.

crh

Set amount to shift chroma-red horizontally.

crv

Set amount to shift chroma-red vertically.

edge

Set edge mode, can be smear, default, or warp.

Commands

This filter supports the all above options as commands.

Display CIE color diagram with pixels overlaid onto it.

The filter accepts the following options:

system

Set color system.

ntsc, 470m

ebu, 470bg

smpte

240m

apple

widergb

cie1931

rec709, hdtv

uhdtv, rec2020

dcip3

cie

Set CIE system.

xyy

ucs

luv

nil

gamuts

Set what gamuts to draw. See system option for available values.

size, s

Set ciescope size, by default set to 512.

intensity, i

Set intensity used to map input pixel values to CIE diagram.

contrast

Set contrast used to draw tongue colors that are out of active color system gamut.

corrgamma

Correct gamma displayed on scope, by default enabled.

showwhite

Show white point on CIE diagram, by default disabled.

gamma

Set input gamma. Used only with XYZ input color space.

Visualize information exported by some codecs.

Some codecs can export information through frames using side-data or other means. For example, some MPEG based codecs export motion vectors through the export_mvs flag in the codec flags2 option.

The filter accepts the following option:

1. Set motion vectors to visualize. Available flags for mv are:

   pf

   forward predicted MVs of P-frames

   bf

   forward predicted MVs of B-frames

   bb

   backward predicted MVs of B-frames

2. Display quantization parameters using the chroma planes.
3. Set motion vectors type to visualize. Includes MVs from all frames unless specified by frame_type option. Available flags for mv_type are:

   fp

   forward predicted MVs

   bp

   backward predicted MVs

4. Set frame type to visualize motion vectors of. Available flags for frame_type are:

   if

   intra-coded frames (I-frames)

   pf

   predicted frames (P-frames)

   bf

   bi-directionally predicted frames (B-frames)

Examples

• Visualize forward predicted MVs of all frames using ffplay: ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv_type=fp
• Visualize multi-directionals MVs of P and B-Frames using ffplay: ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv=pf+bf+bb

Modify intensity of primary colors (red, green and blue) of input frames.

The filter allows an input frame to be adjusted in the shadows, midtones or highlights regions for the red-cyan, green-magenta or blue-yellow balance.

A positive adjustment value shifts the balance towards the primary color, a negative value towards the complementary color.

The filter accepts the following options:

rs

gs

bs

Adjust red, green and blue shadows (darkest pixels).

rm

gm

bm

Adjust red, green and blue midtones (medium pixels).

rh

gh

bh

Adjust red, green and blue highlights (brightest pixels). Allowed ranges for options are [-1.0, 1.0]. Defaults are 0.

pl

Preserve lightness when changing color balance. Default is disabled.

Examples

• Add red color cast to shadows: colorbalance=rs=.3

Commands

This filter supports the all above options as commands.

Adjust color contrast between RGB components.

The filter accepts the following options:

rc

Set the red-cyan contrast. Defaults is 0.0. Allowed range is from -1.0 to 1.0.

gm

Set the green-magenta contrast. Defaults is 0.0. Allowed range is from -1.0 to 1.0.

by

Set the blue-yellow contrast. Defaults is 0.0. Allowed range is from -1.0 to 1.0.

rcw

gmw

byw

Set the weight of each rc, gm, by option value. Default value is 0.0. Allowed range is from 0.0 to 1.0. If all weights are 0.0 filtering is disabled.

pl

Set the amount of preserving lightness. Default value is 0.0. Allowed range is from 0.0 to 1.0.

Commands

This filter supports the all above options as commands.

Adjust color white balance selectively for blacks and whites. This filter operates in YUV colorspace.

The filter accepts the following options:

rl

Set the red shadow spot. Allowed range is from -1.0 to 1.0. Default value is 0.

bl

Set the blue shadow spot. Allowed range is from -1.0 to 1.0. Default value is 0.

rh

Set the red highlight spot. Allowed range is from -1.0 to 1.0. Default value is 0.

bh

Set the red highlight spot. Allowed range is from -1.0 to 1.0. Default value is 0.

saturation

Set the amount of saturation. Allowed range is from -3.0 to 3.0. Default value is 1.

Commands

This filter supports the all above options as commands.

Adjust video input frames by re-mixing color channels.

This filter modifies a color channel by adding the values associated to the other channels of the same pixels. For example if the value to modify is red, the output value will be:

<red>=<red>*<rr> + <blue>*<rb> + <green>*<rg> + <alpha>*<ra>

The filter accepts the following options:

rr

rg

rb

ra

Adjust contribution of input red, green, blue and alpha channels for output red channel. Default is 1 for rr, and 0 for rg, rb and ra.

gr

gg

gb

ga

Adjust contribution of input red, green, blue and alpha channels for output green channel. Default is 1 for gg, and 0 for gr, gb and ga.

br

bg

bb

ba

Adjust contribution of input red, green, blue and alpha channels for output blue channel. Default is 1 for bb, and 0 for br, bg and ba.

ar

ag

ab

aa

Adjust contribution of input red, green, blue and alpha channels for output alpha channel. Default is 1 for aa, and 0 for ar, ag and ab. Allowed ranges for options are [-2.0, 2.0].

pl

Preserve lightness when changing colors. Allowed range is from [0.0, 1.0]. Default is 0.0, thus disabled.

Examples

• Convert source to grayscale: colorchannelmixer=.3:.4:.3:0:.3:.4:.3:0:.3:.4:.3
• Simulate sepia tones: colorchannelmixer=.393:.769:.189:0:.349:.686:.168:0:.272:.534:.131

Commands

This filter supports the all above options as commands.

Overlay a solid color on the video stream.

The filter accepts the following options:

hue

Set the color hue. Allowed range is from 0 to 360. Default value is 0.

saturation

Set the color saturation. Allowed range is from 0 to 1. Default value is 0.5.

lightness

Set the color lightness. Allowed range is from 0 to 1. Default value is 0.5.

(no term)

Set the mix of source lightness. By default is set to 1.0. Allowed range is from 0.0 to 1.0.

Commands

This filter supports the all above options as commands.

RGB colorspace color keying.

The filter accepts the following options:

color

The color which will be replaced with transparency.

similarity

Similarity percentage with the key color. 0.01 matches only the exact key color, while 1.0 matches everything.

blend

Blend percentage. 0.0 makes pixels either fully transparent, or not transparent at all. Higher values result in semi-transparent pixels, with a higher transparency the more similar the pixels color is to the key color.

Examples

• Make every green pixel in the input image transparent: ffmpeg -i input.png -vf colorkey=green out.png
• Overlay a greenscreen-video on top of a static background image. ffmpeg -i background.png -i video.mp4 -filter_complex “[1:v]colorkey=0x3BBD1E:0.3:0.2[ckout];[0:v][ckout]overlay[out]” -map “[out]” output.flv

Commands

This filter supports same commands as options. The command accepts the same syntax of the corresponding option.

If the specified expression is not valid, it is kept at its current value.

Remove all color information for all RGB colors except for certain one.

The filter accepts the following options:

color

The color which will not be replaced with neutral gray.

similarity

Similarity percentage with the above color. 0.01 matches only the exact key color, while 1.0 matches everything.

blend

Blend percentage. 0.0 makes pixels fully gray. Higher values result in more preserved color.

Commands

This filter supports same commands as options. The command accepts the same syntax of the corresponding option.

If the specified expression is not valid, it is kept at its current value.

Adjust video input frames using levels.

The filter accepts the following options:

rimin

gimin

bimin

aimin

Adjust red, green, blue and alpha input black point. Allowed ranges for options are [-1.0, 1.0]. Defaults are 0.

rimax

gimax

bimax

aimax

Adjust red, green, blue and alpha input white point. Allowed ranges for options are [-1.0, 1.0]. Defaults are 1. Input levels are used to lighten highlights (bright tones), darken shadows (dark tones), change the balance of bright and dark tones.

romin

gomin

bomin

aomin

Adjust red, green, blue and alpha output black point. Allowed ranges for options are [0, 1.0]. Defaults are 0.

romax

gomax

bomax

aomax

Adjust red, green, blue and alpha output white point. Allowed ranges for options are [0, 1.0]. Defaults are 1. Output levels allows manual selection of a constrained output level range.

Examples

• Make video output darker: colorlevels=rimin=0.058:gimin=0.058:bimin=0.058
• Increase contrast: colorlevels=rimin=0.039:gimin=0.039:bimin=0.039:rimax=0.96:gimax=0.96:bimax=0.96
• Make video output lighter: colorlevels=rimax=0.902:gimax=0.902:bimax=0.902
• Increase brightness: colorlevels=romin=0.5:gomin=0.5:bomin=0.5

Commands

This filter supports the all above options as commands.

Convert color matrix.

The filter accepts the following options:

src

dst

Specify the source and destination color matrix. Both values must be specified. The accepted values are:

bt709

BT.709

fcc

FCC

bt601

BT.601

bt470

BT.470

bt470bg

BT.470BG

smpte170m

SMPTE-170M

smpte240m

SMPTE-240M

bt2020

BT.2020

nil

For example to convert from BT.601 to SMPTE-240M, use the command:

colormatrix=bt601:smpte240m

Convert colorspace, transfer characteristics or color primaries. Input video needs to have an even size.

The filter accepts the following options:

all

Specify all color properties at once. The accepted values are:

bt470m

BT.470M

bt470bg

BT.470BG

bt601-6-525

BT.601-6 525

bt601-6-625

BT.601-6 625

bt709

BT.709

smpte170m

SMPTE-170M

smpte240m

SMPTE-240M

bt2020

BT.2020

space

Specify output colorspace. The accepted values are:

bt709

BT.709

fcc

FCC

bt470bg

BT.470BG or BT.601-6 625

smpte170m

SMPTE-170M or BT.601-6 525

smpte240m

SMPTE-240M

ycgco

YCgCo

bt2020ncl

BT.2020 with non-constant luminance

trc

Specify output transfer characteristics. The accepted values are:

bt709

BT.709

bt470m

BT.470M

bt470bg

BT.470BG

gamma22

Constant gamma of 2.2

gamma28

Constant gamma of 2.8

smpte170m

SMPTE-170M, BT.601-6 625 or BT.601-6 525

smpte240m

SMPTE-240M

srgb

SRGB

iec61966-2-1

iec61966-2-1

iec61966-2-4

iec61966-2-4

xvycc

xvycc

bt2020-10

BT.2020 for 10-bits content

bt2020-12

BT.2020 for 12-bits content

primaries

Specify output color primaries. The accepted values are:

bt709

BT.709

bt470m

BT.470M

bt470bg

BT.470BG or BT.601-6 625

smpte170m

SMPTE-170M or BT.601-6 525

smpte240m

SMPTE-240M

film

film

smpte431

SMPTE-431

smpte432

SMPTE-432

bt2020

BT.2020

jedec-p22

JEDEC P22 phosphors

range

Specify output color range. The accepted values are:

tv

TV (restricted) range

mpeg

MPEG (restricted) range

pc

PC (full) range

jpeg

JPEG (full) range

format

Specify output color format. The accepted values are:

yuv420p

YUV 4:2:0 planar 8-bits

yuv420p10

YUV 4:2:0 planar 10-bits

yuv420p12

YUV 4:2:0 planar 12-bits

yuv422p

YUV 4:2:2 planar 8-bits

yuv422p10

YUV 4:2:2 planar 10-bits

yuv422p12

YUV 4:2:2 planar 12-bits

yuv444p

YUV 4:4:4 planar 8-bits

yuv444p10

YUV 4:4:4 planar 10-bits

yuv444p12

YUV 4:4:4 planar 12-bits

fast

Do a fast conversion, which skips gamma/primary correction. This will take significantly less CPU, but will be mathematically incorrect. To get output compatible with that produced by the colormatrix filter, use fast=1.

dither

Specify dithering mode. The accepted values are:

none

No dithering

fsb

Floyd-Steinberg dithering

wpadapt

Whitepoint adaptation mode. The accepted values are:

bradford

Bradford whitepoint adaptation

vonkries

von Kries whitepoint adaptation

identity

identity whitepoint adaptation (i.e. no whitepoint adaptation)

iall

Override all input properties at once. Same accepted values as all.

ispace

Override input colorspace. Same accepted values as space.

iprimaries

Override input color primaries. Same accepted values as primaries.

itrc

Override input transfer characteristics. Same accepted values as trc.

irange

Override input color range. Same accepted values as range.

The filter converts the transfer characteristics, color space and color primaries to the specified user values. The output value, if not specified, is set to a default value based on the all property. If that property is also not specified, the filter will log an error. The output color range and format default to the same value as the input color range and format. The input transfer characteristics, color space, color primaries and color range should be set on the input data. If any of these are missing, the filter will log an error and no conversion will take place.

For example to convert the input to SMPTE-240M, use the command:

colorspace=smpte240m

Adjust color temperature in video to simulate variations in ambient color temperature.

The filter accepts the following options:

temperature

Set the temperature in Kelvin. Allowed range is from 1000 to 40000. Default value is 6500 K.

(no term)

Set mixing with filtered output. Allowed range is from 0 to 1. Default value is 1.

pl

Set the amount of preserving lightness. Allowed range is from 0 to 1. Default value is 0.

Commands

This filter supports same commands as options.

Apply convolution of 3x3, 5x5, 7x7 or horizontal/vertical up to 49 elements.

The filter accepts the following options:

0m

1m

2m

3m

Set matrix for each plane. Matrix is sequence of 9, 25 or 49 signed integers in square mode, and from 1 to 49 odd number of signed integers in row mode.

0rdiv

1rdiv

2rdiv

3rdiv

Set multiplier for calculated value for each plane. If unset or 0, it will be sum of all matrix elements.

0bias

1bias

2bias

3bias

Set bias for each plane. This value is added to the result of the multiplication. Useful for making the overall image brighter or darker. Default is 0.0.

0mode

1mode

2mode

3mode

Set matrix mode for each plane. Can be square, row or column. Default is square.

Commands

This filter supports the all above options as commands.

Examples

• Apply sharpen: convolution=“0 -1 0 -1 5 -1 0 -1 0:0 -1 0 -1 5 -1 0 -1 0:0 -1 0 -1 5 -1 0 -1 0:0 -1 0 -1 5 -1 0 -1 0”
• Apply blur: convolution=“1 1 1 1 1 1 1 1 1:1 1 1 1 1 1 1 1 1:1 1 1 1 1 1 1 1 1:1 1 1 1 1 1 1 1 1:1/9:1/9:1/9:1/9”
• Apply edge enhance: convolution=“0 0 0 -1 1 0 0 0 0:0 0 0 -1 1 0 0 0 0:0 0 0 -1 1 0 0 0 0:0 0 0 -1 1 0 0 0 0:5:1:1:1:0:128:128:128”
• Apply edge detect: convolution=“0 1 0 1 -4 1 0 1 0:0 1 0 1 -4 1 0 1 0:0 1 0 1 -4 1 0 1 0:0 1 0 1 -4 1 0 1 0:5:5:5:1:0:128:128:128”
• Apply laplacian edge detector which includes diagonals: convolution=“1 1 1 1 -8 1 1 1 1:1 1 1 1 -8 1 1 1 1:1 1 1 1 -8 1 1 1 1:1 1 1 1 -8 1 1 1 1:5:5:5:1:0:128:128:0”
• Apply emboss: convolution=“-2 -1 0 -1 1 1 0 1 2:-2 -1 0 -1 1 1 0 1 2:-2 -1 0 -1 1 1 0 1 2:-2 -1 0 -1 1 1 0 1 2”

Apply 2D convolution of video stream in frequency domain using second stream as impulse.

The filter accepts the following options:

planes

Set which planes to process.

impulse

Set which impulse video frames will be processed, can be first or all. Default is all.

The convolve filter also supports the framesync options.

Copy the input video source unchanged to the output. This is mainly useful for testing purposes.

Video filtering on GPU using Apple’s CoreImage API on OSX.

Hardware acceleration is based on an OpenGL context. Usually, this means it is processed by video hardware. However, software-based OpenGL implementations exist which means there is no guarantee for hardware processing. It depends on the respective OSX.

There are many filters and image generators provided by Apple that come with a large variety of options. The filter has to be referenced by its name along with its options.

The coreimage filter accepts the following options:

list_filters

List all available filters and generators along with all their respective options as well as possible minimum and maximum values along with the default values. list_filters=true

filter

Specify all filters by their respective name and options. Use list_filters to determine all valid filter names and options. Numerical options are specified by a float value and are automatically clamped to their respective value range. Vector and color options have to be specified by a list of space separated float values. Character escaping has to be done. A special option name default is available to use default options for a filter. It is required to specify either default or at least one of the filter options. All omitted options are used with their default values. The syntax of the filter string is as follows: filter=<NAME>@<OPTION>=<VALUE>[@<OPTION>=<VALUE>][@…][#<NAME>@<OPTION>=<VALUE>[@<OPTION>=<VALUE>][@…]][#…]

output_rect

Specify a rectangle where the output of the filter chain is copied into the input image. It is given by a list of space separated float values: output_rect=x\ y\ width\ height If not given, the output rectangle equals the dimensions of the input image. The output rectangle is automatically cropped at the borders of the input image. Negative values are valid for each component. output_rect=25\ 25\ 100\ 100

Several filters can be chained for successive processing without GPU-HOST transfers allowing for fast processing of complex filter chains. Currently, only filters with zero (generators) or exactly one (filters) input image and one output image are supported. Also, transition filters are not yet usable as intended.

Some filters generate output images with additional padding depending on the respective filter kernel. The padding is automatically removed to ensure the filter output has the same size as the input image.

For image generators, the size of the output image is determined by the previous output image of the filter chain or the input image of the whole filterchain, respectively. The generators do not use the pixel information of this image to generate their output. However, the generated output is blended onto this image, resulting in partial or complete coverage of the output image.

The coreimagesrc video source can be used for generating input images which are directly fed into the filter chain. By using it, providing input images by another video source or an input video is not required.

Examples

• List all filters available: coreimage=list_filters=true
• Use the CIBoxBlur filter with default options to blur an image: coreimage=filter=CIBoxBlur@default
• Use a filter chain with CISepiaTone at default values and CIVignetteEffect with its center at 100x100 and a radius of 50 pixels: coreimage=filter=CIBoxBlur@default#CIVignetteEffect@inputCenter=100\ 100@inputRadius=50
• Use nullsrc and CIQRCodeGenerator to create a QR code for the FFmpeg homepage, given as complete and escaped command-line for Apple’s standard bash shell: ffmpeg -f lavfi -i nullsrc=s=100x100,coreimage=filter=CIQRCodeGenerator@inputMessage=https\\\\\://FFmpeg.org/@inputCorrectionLevel=H -frames:v 1 QRCode.png

Cover a rectangular object

It accepts the following options:

cover

Filepath of the optional cover image, needs to be in yuv420.

mode

Set covering mode. It accepts the following values:

cover

cover it by the supplied image

blur

cover it by interpolating the surrounding pixels

Default value is blur.

Examples

• Cover a rectangular object by the supplied image of a given video using ffmpeg: ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv

Crop the input video to given dimensions.

It accepts the following parameters:

w, out_w

The width of the output video. It defaults to iw. This expression is evaluated only once during the filter configuration, or when the w or out_w command is sent.

h, out_h

The height of the output video. It defaults to ih. This expression is evaluated only once during the filter configuration, or when the h or out_h command is sent.

(no term)

The horizontal position, in the input video, of the left edge of the output video. It defaults to (in_w-out_w)/2. This expression is evaluated per-frame.

(no term)

The vertical position, in the input video, of the top edge of the output video. It defaults to (in_h-out_h)/2. This expression is evaluated per-frame.

keep_aspect

If set to 1 will force the output display aspect ratio to be the same of the input, by changing the output sample aspect ratio. It defaults to 0.

exact

Enable exact cropping. If enabled, subsampled videos will be cropped at exact width/height/x/y as specified and will not be rounded to nearest smaller value. It defaults to 0.

The out_w, out_h, x, y parameters are expressions containing the following constants:

The computed values for x and y. They are evaluated for each new frame.

in_w

in_h

The input width and height.

iw

ih

These are the same as in_w and in_h.

out_w

out_h

The output (cropped) width and height.

ow

oh

These are the same as out_w and out_h.

1. same as iw / ih
2. input sample aspect ratio
3. input display aspect ratio, it is the same as (iw / ih) * sar

horizontal and vertical chroma subsample values. For example for the pixel format yuv422p hsub is 2 and vsub is 1.

1. The number of the input frame, starting from 0.
2. the position in the file of the input frame, NAN if unknown
3. The timestamp expressed in seconds. It’s NAN if the input timestamp is unknown.

The expression for out_w may depend on the value of out_h, and the expression for out_h may depend on out_w, but they cannot depend on x and y, as x and y are evaluated after out_w and out_h.

The x and y parameters specify the expressions for the position of the top-left corner of the output (non-cropped) area. They are evaluated for each frame. If the evaluated value is not valid, it is approximated to the nearest valid value.

The expression for x may depend on y, and the expression for y may depend on x.

Examples

• Crop area with size 100x100 at position (12,34). crop=100:100:12:34 Using named options, the example above becomes: crop=w=100:h=100:x=12:y=34
• Crop the central input area with size 100x100: crop=100:100
• Crop the central input area with size 2/3 of the input video: crop=2/3*in_w:2/3*in_h
• Crop the input video central square: crop=out_w=in_h crop=in_h
• Delimit the rectangle with the top-left corner placed at position 100:100 and the right-bottom corner corresponding to the right-bottom corner of the input image. crop=in_w-100:in_h-100:100:100
• Crop 10 pixels from the left and right borders, and 20 pixels from the top and bottom borders crop=in_w-2*10:in_h-2*20
• Keep only the bottom right quarter of the input image: crop=in_w/2:in_h/2:in_w/2:in_h/2
• Crop height for getting Greek harmony: crop=in_w:1/PHI*in_w
• Apply trembling effect: crop=in_w/2:in_h/2:(in_w-out_w)/2+((in_w-out_w)/2)*sin(n/10):(in_h-out_h)/2 +((in_h-out_h)/2)*sin(n/7)
• Apply erratic camera effect depending on timestamp: crop=in_w/2:in_h/2:(in_w-out_w)/2+((in_w-out_w)/2)*sin(t*10):(in_h-out_h)/2 +((in_h-out_h)/2)*sin(t*13)“
• Set x depending on the value of y: crop=in_w/2:in_h/2:y:10+10*sin(n/10)

Commands

This filter supports the following commands:

w, out_w

h, out_h

(no term)

(no term)

Set width/height of the output video and the horizontal/vertical position in the input video. The command accepts the same syntax of the corresponding option. If the specified expression is not valid, it is kept at its current value.

Auto-detect the crop size.

It calculates the necessary cropping parameters and prints the recommended parameters via the logging system. The detected dimensions correspond to the non-black area of the input video.

It accepts the following parameters:

limit

Set higher black value threshold, which can be optionally specified from nothing (0) to everything (255 for 8-bit based formats). An intensity value greater to the set value is considered non-black. It defaults to 24. You can also specify a value between 0.0 and 1.0 which will be scaled depending on the bitdepth of the pixel format.

round

The value which the width/height should be divisible by. It defaults to 16. The offset is automatically adjusted to center the video. Use 2 to get only even dimensions (needed for 4:2:2 video). 16 is best when encoding to most video codecs.

skip

Set the number of initial frames for which evaluation is skipped. Default is 2. Range is 0 to INT_MAX.

reset_count, reset

Set the counter that determines after how many frames cropdetect will reset the previously detected largest video area and start over to detect the current optimal crop area. Default value is 0. This can be useful when channel logos distort the video area. 0 indicates ’never reset’, and returns the largest area encountered during playback.

Delay video filtering until a given wallclock timestamp. The filter first passes on preroll amount of frames, then it buffers at most buffer amount of frames and waits for the cue. After reaching the cue it forwards the buffered frames and also any subsequent frames coming in its input.

The filter can be used synchronize the output of multiple ffmpeg processes for realtime output devices like decklink. By putting the delay in the filtering chain and pre-buffering frames the process can pass on data to output almost immediately after the target wallclock timestamp is reached.

Perfect frame accuracy cannot be guaranteed, but the result is good enough for some use cases.

cue

The cue timestamp expressed in a UNIX timestamp in microseconds. Default is 0.

preroll

The duration of content to pass on as preroll expressed in seconds. Default is 0.

buffer

The maximum duration of content to buffer before waiting for the cue expressed in seconds. Default is 0.

Apply color adjustments using curves.

This filter is similar to the Adobe Photoshop and GIMP curves tools. Each component (red, green and blue) has its values defined by N key points tied from each other using a smooth curve. The x-axis represents the pixel values from the input frame, and the y-axis the new pixel values to be set for the output frame.

By default, a component curve is defined by the two points (0;0) and (1;1). This creates a straight line where each original pixel value is adjusted to its own value, which means no change to the image.

The filter allows you to redefine these two points and add some more. A new curve (using a natural cubic spline interpolation) will be define to pass smoothly through all these new coordinates. The new defined points needs to be strictly increasing over the x-axis, and their x and y values must be in the [0;1] interval. If the computed curves happened to go outside the vector spaces, the values will be clipped accordingly.

The filter accepts the following options:

preset

Select one of the available color presets. This option can be used in addition to the r, g, b parameters; in this case, the later options takes priority on the preset values. Available presets are:

none

color_negative

cross_process

darker

increase_contrast

lighter

linear_contrast

medium_contrast

negative

strong_contrast

vintage

Default is none.

master, m

Set the master key points. These points will define a second pass mapping. It is sometimes called a luminance or value mapping. It can be used with r, g, b or all since it acts like a post-processing LUT.

red, r

Set the key points for the red component.

green, g

Set the key points for the green component.

blue, b

Set the key points for the blue component.

all

Set the key points for all components (not including master). Can be used in addition to the other key points component options. In this case, the unset component(s) will fallback on this all setting.

psfile

Specify a Photoshop curves file (.acv) to import the settings from.

plot

Save Gnuplot script of the curves in specified file.

To avoid some filtergraph syntax conflicts, each key points list need to be defined using the following syntax: x0/y0 x1/y1 x2/y2 ....

Commands

This filter supports same commands as options.

Examples

• Increase slightly the middle level of blue: curves=blue=0/0 0.5/0.58 1/1
• Vintage effect: curves=r=0/0.11 .42/.51 1/0.95:g=0/0 0.50/0.48 1/1:b=0/0.22 .49/.44 1/0.8 Here we obtain the following coordinates for each components:

  red

  (0;0.11) (0.42;0.51) (1;0.95)

  green

  (0;0) (0.50;0.48) (1;1)

  blue

  (0;0.22) (0.49;0.44) (1;0.80)

• The previous example can also be achieved with the associated built-in preset: curves=preset=vintage
• Or simply: curves=vintage
• Use a Photoshop preset and redefine the points of the green component: curves=psfile=MyCurvesPresets/purple.acv:green=0/0 0.45/0.53 1/1
• Check out the curves of the cross_process profile using ffmpeg and gnuplot: ffmpeg -f lavfi -i color -vf curves=cross_process:plot=/tmp/curves.plt -frames:v 1 -f null - gnuplot -p /tmp/curves.plt

Video data analysis filter.

This filter shows hexadecimal pixel values of part of video.

The filter accepts the following options:

size, s

Set output video size.

(no term)

Set x offset from where to pick pixels.

(no term)

Set y offset from where to pick pixels.

mode

Set scope mode, can be one of the following:

mono

Draw hexadecimal pixel values with white color on black background.

color

Draw hexadecimal pixel values with input video pixel color on black background.

color2

Draw hexadecimal pixel values on color background picked from input video, the text color is picked in such way so its always visible.

axis

Draw rows and columns numbers on left and top of video.

opacity

Set background opacity.

format

Set display number format. Can be hex, or dec. Default is hex.

components

Set pixel components to display. By default all pixel components are displayed.

Commands

This filter supports same commands as options excluding size option.

Apply Directional blur filter.

The filter accepts the following options:

angle

Set angle of directional blur. Default is 45.

radius

Set radius of directional blur. Default is 5.

planes

Set which planes to filter. By default all planes are filtered.

Commands

This filter supports same commands as options. The command accepts the same syntax of the corresponding option.

If the specified expression is not valid, it is kept at its current value.

Denoise frames using 2D DCT (frequency domain filtering).

This filter is not designed for real time.

The filter accepts the following options:

sigma, s

Set the noise sigma constant. This sigma defines a hard threshold of 3 * sigma; every DCT coefficient (absolute value) below this threshold with be dropped. If you need a more advanced filtering, see expr. Default is 0.

overlap

Set number overlapping pixels for each block. Since the filter can be slow, you may want to reduce this value, at the cost of a less effective filter and the risk of various artefacts. If the overlapping value doesn’t permit processing the whole input width or height, a warning will be displayed and according borders won’t be denoised. Default value is blocksize-1, which is the best possible setting.

expr, e

Set the coefficient factor expression. For each coefficient of a DCT block, this expression will be evaluated as a multiplier value for the coefficient. If this is option is set, the sigma option will be ignored. The absolute value of the coefficient can be accessed through the c variable.

(no term)

Set the blocksize using the number of bits. =1<<=/=n=/ defines the blocksize, which is the width and height of the processed blocks. The default value is 3 (8x8) and can be raised to 4 for a blocksize of 16x16. Note that changing this setting has huge consequences on the speed processing. Also, a larger block size does not necessarily means a better de-noising.

Examples

Apply a denoise with a sigma of 4.5:

dctdnoiz=4.5

The same operation can be achieved using the expression system:

dctdnoiz=e=gte(c, 4.5*3)

Violent denoise using a block size of 16x16:

dctdnoiz=15:n=4

Remove banding artifacts from input video. It works by replacing banded pixels with average value of referenced pixels.

The filter accepts the following options:

1thr

2thr

3thr

4thr

Set banding detection threshold for each plane. Default is 0.02. Valid range is 0.00003 to 0.5. If difference between current pixel and reference pixel is less than threshold, it will be considered as banded.

range, r

Banding detection range in pixels. Default is 16. If positive, random number in range 0 to set value will be used. If negative, exact absolute value will be used. The range defines square of four pixels around current pixel.

direction, d

Set direction in radians from which four pixel will be compared. If positive, random direction from 0 to set direction will be picked. If negative, exact of absolute value will be picked. For example direction 0, -PI or -2*PI radians will pick only pixels on same row and -PI/2 will pick only pixels on same column.

blur, b

If enabled, current pixel is compared with average value of all four surrounding pixels. The default is enabled. If disabled current pixel is compared with all four surrounding pixels. The pixel is considered banded if only all four differences with surrounding pixels are less than threshold.

coupling, c

If enabled, current pixel is changed if and only if all pixel components are banded, e.g. banding detection threshold is triggered for all color components. The default is disabled.

Commands

This filter supports the all above options as commands.

Remove blocking artifacts from input video.

The filter accepts the following options:

filter

Set filter type, can be weak or strong. Default is strong. This controls what kind of deblocking is applied.

block

Set size of block, allowed range is from 4 to 512. Default is 8.

alpha

beta

gamma

delta

Set blocking detection thresholds. Allowed range is 0 to 1. Defaults are: 0.098 for alpha and 0.05 for the rest. Using higher threshold gives more deblocking strength. Setting alpha controls threshold detection at exact edge of block. Remaining options controls threshold detection near the edge. Each one for below/above or left/right. Setting any of those to 0 disables deblocking.

planes

Set planes to filter. Default is to filter all available planes.

Examples

• Deblock using weak filter and block size of 4 pixels. deblock=filter=weak:block=4
• Deblock using strong filter, block size of 4 pixels and custom thresholds for deblocking more edges. deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05
• Similar as above, but filter only first plane. deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=1
• Similar as above, but filter only second and third plane. deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=6

Commands

This filter supports the all above options as commands.

Drop duplicated frames at regular intervals.

The filter accepts the following options:

cycle

Set the number of frames from which one will be dropped. Setting this to N means one frame in every batch of N frames will be dropped. Default is 5.

dupthresh

Set the threshold for duplicate detection. If the difference metric for a frame is less than or equal to this value, then it is declared as duplicate. Default is 1.1

scthresh

Set scene change threshold. Default is 15.

blockx

blocky

Set the size of the x and y-axis blocks used during metric calculations. Larger blocks give better noise suppression, but also give worse detection of small movements. Must be a power of two. Default is 32.

ppsrc

Mark main input as a pre-processed input and activate clean source input stream. This allows the input to be pre-processed with various filters to help the metrics calculation while keeping the frame selection lossless. When set to 1, the first stream is for the pre-processed input, and the second stream is the clean source from where the kept frames are chosen. Default is 0.

chroma

Set whether or not chroma is considered in the metric calculations. Default is 1.

Apply 2D deconvolution of video stream in frequency domain using second stream as impulse.

The filter accepts the following options:

planes

Set which planes to process.

impulse

Set which impulse video frames will be processed, can be first or all. Default is all.

noise

Set noise when doing divisions. Default is 0.0000001. Useful when width and height are not same and not power of 2 or if stream prior to convolving had noise.

The deconvolve filter also supports the framesync options.

Reduce cross-luminance (dot-crawl) and cross-color (rainbows) from video.

It accepts the following options:

1. Set mode of operation. Can be combination of dotcrawl for cross-luminance reduction and/or rainbows for cross-color reduction.
2. Set spatial luma threshold. Lower values increases reduction of cross-luminance.
3. Set tolerance for temporal luma. Higher values increases reduction of cross-luminance.
4. Set tolerance for chroma temporal variation. Higher values increases reduction of cross-color.
5. Set temporal chroma threshold. Lower values increases reduction of cross-color.

Apply deflate effect to the video.

This filter replaces the pixel by the local(3x3) average by taking into account only values lower than the pixel.

It accepts the following options:

threshold0

threshold1

threshold2

threshold3

Limit the maximum change for each plane, default is 65535. If 0, plane will remain unchanged.

Commands

This filter supports the all above options as commands.

Remove temporal frame luminance variations.

It accepts the following options:

size, s

Set moving-average filter size in frames. Default is 5. Allowed range is 2 - 129.

mode, m

Set averaging mode to smooth temporal luminance variations. Available values are:

am

Arithmetic mean

gm

Geometric mean

hm

Harmonic mean

qm

Quadratic mean

(no term)

Cubic mean

pm

Power mean

median

Median

bypass

Do not actually modify frame. Useful when one only wants metadata.

Remove judder produced by partially interlaced telecined content.

Judder can be introduced, for instance, by pullup filter. If the original source was partially telecined content then the output of pullup,dejudder will have a variable frame rate. May change the recorded frame rate of the container. Aside from that change, this filter will not affect constant frame rate video.

The option available in this filter is:

cycle

Specify the length of the window over which the judder repeats. Accepts any integer greater than 1. Useful values are:

1. If the original was telecined from 24 to 30 fps (Film to NTSC).
2. If the original was telecined from 25 to 30 fps (PAL to NTSC).
3. If a mixture of the two.

The default is 4.

Suppress a TV station logo by a simple interpolation of the surrounding pixels. Just set a rectangle covering the logo and watch it disappear (and sometimes something even uglier appear - your mileage may vary).

It accepts the following parameters:

Specify the top left corner coordinates of the logo. They must be specified.

Specify the width and height of the logo to clear. They must be specified.

show

When set to 1, a green rectangle is drawn on the screen to simplify finding the right x, y, w, and h parameters. The default value is 0. The rectangle is drawn on the outermost pixels which will be (partly) replaced with interpolated values. The values of the next pixels immediately outside this rectangle in each direction will be used to compute the interpolated pixel values inside the rectangle.

Examples

• Set a rectangle covering the area with top left corner coordinates 0,0 and size 100x77: delogo=x=0:y=0:w=100:h=77

Remove the rain in the input image/video by applying the derain methods based on convolutional neural networks. Supported models:

• Recurrent Squeeze-and-Excitation Context Aggregation Net (RESCAN). See <*http://openaccess.thecvf.com/content_ECCV_2018/papers/Xia_Li_Recurrent_Squeeze-and-Excitation_Context_ECCV_2018_paper.pdf*>.

Training as well as model generation scripts are provided in the repository at <*https://github.com/XueweiMeng/derain_filter.git*>.

Native model files (.model) can be generated from TensorFlow model files (.pb) by using tools/python/convert.py

The filter accepts the following options:

filter_type

Specify which filter to use. This option accepts the following values:

derain

Derain filter. To conduct derain filter, you need to use a derain model.

dehaze

Dehaze filter. To conduct dehaze filter, you need to use a dehaze model.

Default value is derain.

dnn_backend

Specify which DNN backend to use for model loading and execution. This option accepts the following values:

native

Native implementation of DNN loading and execution.

tensorflow

TensorFlow backend. To enable this backend you need to install the TensorFlow for C library (see <*https://www.tensorflow.org/install/install_c*>) and configure FFmpeg with --enable-libtensorflow

Default value is native.

model

Set path to model file specifying network architecture and its parameters. Note that different backends use different file formats. TensorFlow and native backend can load files for only its format.

It can also be finished with dnn_processing filter.

Attempt to fix small changes in horizontal and/or vertical shift. This filter helps remove camera shake from hand-holding a camera, bumping a tripod, moving on a vehicle, etc.

The filter accepts the following options:

Specify a rectangular area where to limit the search for motion vectors. If desired the search for motion vectors can be limited to a rectangular area of the frame defined by its top left corner, width and height. These parameters have the same meaning as the drawbox filter which can be used to visualise the position of the bounding box. This is useful when simultaneous movement of subjects within the frame might be confused for camera motion by the motion vector search. If any or all of x, y, w and h are set to -1 then the full frame is used. This allows later options to be set without specifying the bounding box for the motion vector search. Default - search the whole frame.

rx

ry

Specify the maximum extent of movement in x and y directions in the range 0-64 pixels. Default 16.

edge

Specify how to generate pixels to fill blanks at the edge of the frame. Available values are:

blank, 0

Fill zeroes at blank locations

original, 1

Original image at blank locations

clamp, 2

Extruded edge value at blank locations

mirror, 3

Mirrored edge at blank locations

Default value is mirror.

blocksize

Specify the blocksize to use for motion search. Range 4-128 pixels, default 8.

contrast

Specify the contrast threshold for blocks. Only blocks with more than the specified contrast (difference between darkest and lightest pixels) will be considered. Range 1-255, default 125.

search

Specify the search strategy. Available values are:

exhaustive, 0

Set exhaustive search

less, 1

Set less exhaustive search.

Default value is exhaustive.

filename

If set then a detailed log of the motion search is written to the specified file.

Remove unwanted contamination of foreground colors, caused by reflected color of greenscreen or bluescreen.

This filter accepts the following options:

type

Set what type of despill to use.

(no term)

Set how spillmap will be generated.

expand

Set how much to get rid of still remaining spill.

red

Controls amount of red in spill area.

green

Controls amount of green in spill area. Should be -1 for greenscreen.

blue

Controls amount of blue in spill area. Should be -1 for bluescreen.

brightness

Controls brightness of spill area, preserving colors.

alpha

Modify alpha from generated spillmap.

Commands

This filter supports the all above options as commands.

Apply an exact inverse of the telecine operation. It requires a predefined pattern specified using the pattern option which must be the same as that passed to the telecine filter.

This filter accepts the following options:

first_field

top, t

top field first

bottom, b

bottom field first The default value is top.

pattern

A string of numbers representing the pulldown pattern you wish to apply. The default value is 23.

start_frame

A number representing position of the first frame with respect to the telecine pattern. This is to be used if the stream is cut. The default value is 0.

Apply dilation effect to the video.

This filter replaces the pixel by the local(3x3) maximum.

It accepts the following options:

threshold0

threshold1

threshold2

threshold3

Limit the maximum change for each plane, default is 65535. If 0, plane will remain unchanged.

coordinates

Flag which specifies the pixel to refer to. Default is 255 i.e. all eight pixels are used. Flags to local 3x3 coordinates maps like this: 1 2 3 4 5 6 7 8

Commands

This filter supports the all above options as commands.

Displace pixels as indicated by second and third input stream.

It takes three input streams and outputs one stream, the first input is the source, and second and third input are displacement maps.

The second input specifies how much to displace pixels along the x-axis, while the third input specifies how much to displace pixels along the y-axis. If one of displacement map streams terminates, last frame from that displacement map will be used.

Note that once generated, displacements maps can be reused over and over again.

A description of the accepted options follows.

edge

Set displace behavior for pixels that are out of range. Available values are:

blank

Missing pixels are replaced by black pixels.

smear

Adjacent pixels will spread out to replace missing pixels.

wrap

Out of range pixels are wrapped so they point to pixels of other side.

mirror

Out of range pixels will be replaced with mirrored pixels.

Default is smear.

Examples

• Add ripple effect to rgb input of video size hd720: ffmpeg -i INPUT -f lavfi -i nullsrc=s=hd720,lutrgb=128:128:128 -f lavfi -i nullsrc=s=hd720,geq=r=128+30*sin(2*PI*X/400+T):g=128+30*sin(2*PI*X/400+T):b=128+30*sin(2*PI*X/400+T) -lavfi [0][1][2]displace OUTPUT
• Add wave effect to rgb input of video size hd720: ffmpeg -i INPUT -f lavfi -i nullsrc=hd720,geq=r=128+80*(sin(sqrt((X-W/2)*(X-W/2)+(Y-H/2)*(Y-H/2))/220*2*PI+T)):g=128+80*(sin(sqrt((X-W/2)*(X-W/2)+(Y-H/2)*(Y-H/2))/220*2*PI+T)):b=128+80*(sin(sqrt((X-W/2)*(X-W/2)+(Y-H/2)*(Y-H/2))/220*2*PI+T)) -lavfi [1]split[x][y],[0][x][y]displace OUTPUT

Do image processing with deep neural networks. It works together with another filter which converts the pixel format of the Frame to what the dnn network requires.

The filter accepts the following options:

dnn_backend

Specify which DNN backend to use for model loading and execution. This option accepts the following values:

native

Native implementation of DNN loading and execution.

tensorflow

TensorFlow backend. To enable this backend you need to install the TensorFlow for C library (see <*https://www.tensorflow.org/install/install_c*>) and configure FFmpeg with --enable-libtensorflow

openvino

OpenVINO backend. To enable this backend you need to build and install the OpenVINO for C library (see <*https://github.com/openvinotoolkit/openvino/blob/master/build-instruction.md*>) and configure FFmpeg with --enable-libopenvino (–extra-cflags=-I… –extra-ldflags=-L… might be needed if the header files and libraries are not installed into system path)

Default value is native.

model

Set path to model file specifying network architecture and its parameters. Note that different backends use different file formats. TensorFlow, OpenVINO and native backend can load files for only its format. Native model file (.model) can be generated from TensorFlow model file (.pb) by using tools/python/convert.py

input

Set the input name of the dnn network.

output

Set the output name of the dnn network.

async

use DNN async execution if set (default: set), roll back to sync execution if the backend does not support async.

Examples

• Remove rain in rgb24 frame with can.pb (see derain filter): ./ffmpeg -i rain.jpg -vf format=rgb24,dnn_processing=dnn_backend=tensorflow:model=can.pb:input=x:output=y derain.jpg
• Halve the pixel value of the frame with format gray32f: ffmpeg -i input.jpg -vf format=grayf32,dnn_processing=model=halve_gray_float.model:input=dnn_in:output=dnn_out:dnn_backend=native -y out.native.png
• Handle the Y channel with srcnn.pb (see sr filter) for frame with yuv420p (planar YUV formats supported): ./ffmpeg -i 480p.jpg -vf format=yuv420p,scale=w=iw*2:h=ih*2,dnn_processing=dnn_backend=tensorflow:model=srcnn.pb:input=x:output=y -y srcnn.jpg
• Handle the Y channel with espcn.pb (see sr filter), which changes frame size, for format yuv420p (planar YUV formats supported): ./ffmpeg -i 480p.jpg -vf format=yuv420p,dnn_processing=dnn_backend=tensorflow:model=espcn.pb:input=x:output=y -y tmp.espcn.jpg

Draw a colored box on the input image.

It accepts the following parameters:

The expressions which specify the top left corner coordinates of the box. It defaults to 0.

width, w

height, h

The expressions which specify the width and height of the box; if 0 they are interpreted as the input width and height. It defaults to 0.

color, c

Specify the color of the box to write. For the general syntax of this option, check the Color section in the ffmpeg-utils manual. If the special value invert is used, the box edge color is the same as the video with inverted luma.

thickness, t

The expression which sets the thickness of the box edge. A value of fill will create a filled box. Default value is 3. See below for the list of accepted constants.

replace

Applicable if the input has alpha. With value 1, the pixels of the painted box will overwrite the video’s color and alpha pixels. Default is 0, which composites the box onto the input, leaving the video’s alpha intact.

The parameters for x, y, w and h and t are expressions containing the following constants:

dar

The input display aspect ratio, it is the same as (w / h) * sar.

hsub

vsub

horizontal and vertical chroma subsample values. For example for the pixel format yuv422p hsub is 2 and vsub is 1.

in_h, ih

in_w, iw

The input width and height.

sar

The input sample aspect ratio.

(no term)

(no term)

The x and y offset coordinates where the box is drawn.

The width and height of the drawn box.

1. The thickness of the drawn box. These constants allow the x, y, w, h and t expressions to refer to each other, so you may for example specify y=x/dar or h=w/dar.

Examples

• Draw a black box around the edge of the input image: drawbox
• Draw a box with color red and an opacity of 50%: drawbox=10:20:200:60:red@0.5 The previous example can be specified as: drawbox=x=10:y=20:w=200:h=60:color=red@0.5
• Fill the box with pink color: drawbox=x=10:y=10:w=100:h=100:color=pink@0.5:t=fill
• Draw a 2-pixel red 2.40:1 mask: drawbox=x=-t:y=0.5*(ih-iw/2.4)-t:w=iw+t*2:h=iw/2.4+t*2:t=2:c=red

Commands

This filter supports same commands as options. The command accepts the same syntax of the corresponding option.

If the specified expression is not valid, it is kept at its current value.

Draw a graph using input video metadata.

It accepts the following parameters:

m1

Set 1st frame metadata key from which metadata values will be used to draw a graph.

fg1

Set 1st foreground color expression.

m2

Set 2nd frame metadata key from which metadata values will be used to draw a graph.

fg2

Set 2nd foreground color expression.

m3

Set 3rd frame metadata key from which metadata values will be used to draw a graph.

fg3

Set 3rd foreground color expression.

m4

Set 4th frame metadata key from which metadata values will be used to draw a graph.

fg4

Set 4th foreground color expression.

min

Set minimal value of metadata value.

max

Set maximal value of metadata value.

bg

Set graph background color. Default is white.

mode

Set graph mode. Available values for mode is:

bar

dot

line

Default is line.

slide

Set slide mode. Available values for slide is:

frame

Draw new frame when right border is reached.

replace

Replace old columns with new ones.

scroll

Scroll from right to left.

rscroll

Scroll from left to right.

picture

Draw single picture.

Default is frame.

size

Set size of graph video. For the syntax of this option, check the Video size section in the ffmpeg-utils manual. The default value is 900x256.

rate, r

Set the output frame rate. Default value is 25. The foreground color expressions can use the following variables:

MIN

Minimal value of metadata value.

MAX

Maximal value of metadata value.

VAL

Current metadata key value.

The color is defined as 0xAABBGGRR.

Example using metadata from signalstats filter:

signalstats,drawgraph=lavfi.signalstats.YAVG:min=0:max=255

Example using metadata from ebur128 filter:

ebur128=metadata=1,adrawgraph=lavfi.r128.M:min=-120:max=5

Draw a grid on the input image.

It accepts the following parameters:

The expressions which specify the coordinates of some point of grid intersection (meant to configure offset). Both default to 0.

width, w

height, h

The expressions which specify the width and height of the grid cell, if 0 they are interpreted as the input width and height, respectively, minus thickness, so image gets framed. Default to 0.

color, c

Specify the color of the grid. For the general syntax of this option, check the Color section in the ffmpeg-utils manual. If the special value invert is used, the grid color is the same as the video with inverted luma.

thickness, t

The expression which sets the thickness of the grid line. Default value is 1. See below for the list of accepted constants.

replace

Applicable if the input has alpha. With 1 the pixels of the painted grid will overwrite the video’s color and alpha pixels. Default is 0, which composites the grid onto the input, leaving the video’s alpha intact.

The parameters for x, y, w and h and t are expressions containing the following constants:

dar

The input display aspect ratio, it is the same as (w / h) * sar.

hsub

vsub

horizontal and vertical chroma subsample values. For example for the pixel format yuv422p hsub is 2 and vsub is 1.

in_h, ih

in_w, iw

The input grid cell width and height.

sar

The input sample aspect ratio.

(no term)

(no term)

The x and y coordinates of some point of grid intersection (meant to configure offset).

The width and height of the drawn cell.

1. The thickness of the drawn cell. These constants allow the x, y, w, h and t expressions to refer to each other, so you may for example specify y=x/dar or h=w/dar.

Examples

• Draw a grid with cell 100x100 pixels, thickness 2 pixels, with color red and an opacity of 50%: drawgrid=width=100:height=100:thickness=2:color=red@0.5
• Draw a white 3x3 grid with an opacity of 50%: drawgrid=w=iw/3:h=ih/3:t=2:c=white@0.5

Commands

This filter supports same commands as options. The command accepts the same syntax of the corresponding option.

If the specified expression is not valid, it is kept at its current value.

Draw a text string or text from a specified file on top of a video, using the libfreetype library.

To enable compilation of this filter, you need to configure FFmpeg with --enable-libfreetype. To enable default font fallback and the font option you need to configure FFmpeg with --enable-libfontconfig. To enable the text_shaping option, you need to configure FFmpeg with --enable-libfribidi.

Syntax

It accepts the following parameters:

box

Used to draw a box around text using the background color. The value must be either 1 (enable) or 0 (disable). The default value of box is 0.

boxborderw

Set the width of the border to be drawn around the box using boxcolor. The default value of boxborderw is 0.

boxcolor

The color to be used for drawing box around text. For the syntax of this option, check the Color section in the ffmpeg-utils manual. The default value of boxcolor is white.

line_spacing

Set the line spacing in pixels of the border to be drawn around the box using box. The default value of line_spacing is 0.

borderw

Set the width of the border to be drawn around the text using bordercolor. The default value of borderw is 0.

bordercolor

Set the color to be used for drawing border around text. For the syntax of this option, check the Color section in the ffmpeg-utils manual. The default value of bordercolor is black.

expansion

Select how the text is expanded. Can be either none, strftime (deprecated) or normal (default). See the drawtext_expansion, Text expansion section below for details.

basetime

Set a start time for the count. Value is in microseconds. Only applied in the deprecated strftime expansion mode. To emulate in normal expansion mode use the pts function, supplying the start time (in seconds) as the second argument.

fix_bounds

If true, check and fix text coords to avoid clipping.

fontcolor

The color to be used for drawing fonts. For the syntax of this option, check the Color section in the ffmpeg-utils manual. The default value of fontcolor is black.

fontcolor_expr

String which is expanded the same way as text to obtain dynamic fontcolor value. By default this option has empty value and is not processed. When this option is set, it overrides fontcolor option.

font

The font family to be used for drawing text. By default Sans.

fontfile

The font file to be used for drawing text. The path must be included. This parameter is mandatory if the fontconfig support is disabled.

alpha

Draw the text applying alpha blending. The value can be a number between 0.0 and 1.0. The expression accepts the same variables x, y as well. The default value is 1. Please see fontcolor_expr.

fontsize

The font size to be used for drawing text. The default value of fontsize is 16.

text_shaping

If set to 1, attempt to shape the text (for example, reverse the order of right-to-left text and join Arabic characters) before drawing it. Otherwise, just draw the text exactly as given. By default 1 (if supported).

ft_load_flags

The flags to be used for loading the fonts. The flags map the corresponding flags supported by libfreetype, and are a combination of the following values:

default

no_scale

no_hinting

render

no_bitmap

vertical_layout

force_autohint

crop_bitmap

pedantic

ignore_global_advance_width

no_recurse

ignore_transform

monochrome

linear_design

no_autohint

Default value is default. For more information consult the documentation for the FT_LOAD_* libfreetype flags.

shadowcolor

The color to be used for drawing a shadow behind the drawn text. For the syntax of this option, check the Color section in the ffmpeg-utils manual. The default value of shadowcolor is black.

shadowx

shadowy

The x and y offsets for the text shadow position with respect to the position of the text. They can be either positive or negative values. The default value for both is 0.

start_number

The starting frame number for the n/frame_num variable. The default value is 0.

tabsize

The size in number of spaces to use for rendering the tab. Default value is 4.

timecode

Set the initial timecode representation in hh:mm:ss[:;.]ff format. It can be used with or without text parameter. timecode_rate option must be specified.

timecode_rate, rate, r

Set the timecode frame rate (timecode only). Value will be rounded to nearest integer. Minimum value is 1. Drop-frame timecode is supported for frame rates 30 & 60.

tc24hmax

If set to 1, the output of the timecode option will wrap around at 24 hours. Default is 0 (disabled).

text

The text string to be drawn. The text must be a sequence of UTF-8 encoded characters. This parameter is mandatory if no file is specified with the parameter textfile.

textfile

A text file containing text to be drawn. The text must be a sequence of UTF-8 encoded characters. This parameter is mandatory if no text string is specified with the parameter text. If both text and textfile are specified, an error is thrown.

reload

If set to 1, the textfile will be reloaded before each frame. Be sure to update it atomically, or it may be read partially, or even fail.

(no term)

(no term)

The expressions which specify the offsets where text will be drawn within the video frame. They are relative to the top/left border of the output image. The default value of x and y is 0. See below for the list of accepted constants and functions.

The parameters for x and y are expressions containing the following constants and functions:

dar

input display aspect ratio, it is the same as (w / h) * sar

hsub

vsub

horizontal and vertical chroma subsample values. For example for the pixel format yuv422p hsub is 2 and vsub is 1.

line_h, lh

the height of each text line

main_h, h, H

the input height

main_w, w, W

the input width

max_glyph_a, ascent

the maximum distance from the baseline to the highest/upper grid coordinate used to place a glyph outline point, for all the rendered glyphs. It is a positive value, due to the grid’s orientation with the Y axis upwards.

max_glyph_d, descent

the maximum distance from the baseline to the lowest grid coordinate used to place a glyph outline point, for all the rendered glyphs. This is a negative value, due to the grid’s orientation, with the Y axis upwards.

max_glyph_h

maximum glyph height, that is the maximum height for all the glyphs contained in the rendered text, it is equivalent to ascent - descent.

max_glyph_w

maximum glyph width, that is the maximum width for all the glyphs contained in the rendered text

(no term)

the number of input frame, starting from 0

rand(min, max)

return a random number included between min and max

sar

The input sample aspect ratio.

(no term)

timestamp expressed in seconds, NAN if the input timestamp is unknown

text_h, th

the height of the rendered text

text_w, tw

the width of the rendered text

(no term)

(no term)

the x and y offset coordinates where the text is drawn. These parameters allow the x and y expressions to refer to each other, so you can for example specify y=x/dar.

pict_type

A one character description of the current frame’s picture type.

pkt_pos

The current packet’s position in the input file or stream (in bytes, from the start of the input). A value of -1 indicates this info is not available.

pkt_duration

The current packet’s duration, in seconds.

pkt_size

The current packet’s size (in bytes).

Text expansion

If expansion is set to strftime, the filter recognizes strftime() sequences in the provided text and expands them accordingly. Check the documentation of strftime(). This feature is deprecated.

If expansion is set to none, the text is printed verbatim.

If expansion is set to normal (which is the default), the following expansion mechanism is used.

The backslash character \, followed by any character, always expands to the second character.

Sequences of the form %{...} are expanded. The text between the braces is a function name, possibly followed by arguments separated by ’:’. If the arguments contain special characters or delimiters (’:’ or ’}’), they should be escaped.

Note that they probably must also be escaped as the value for the text option in the filter argument string and as the filter argument in the filtergraph description, and possibly also for the shell, that makes up to four levels of escaping; using a text file avoids these problems.

The following functions are available:

expr, e

The expression evaluation result. It must take one argument specifying the expression to be evaluated, which accepts the same constants and functions as the x and y values. Note that not all constants should be used, for example the text size is not known when evaluating the expression, so the constants text_w and text_h will have an undefined value.

expr_int_format, eif

Evaluate the expression’s value and output as formatted integer. The first argument is the expression to be evaluated, just as for the expr function. The second argument specifies the output format. Allowed values are x, X, d and u. They are treated exactly as in the printf function. The third parameter is optional and sets the number of positions taken by the output. It can be used to add padding with zeros from the left.

gmtime

The time at which the filter is running, expressed in UTC. It can accept an argument: a strftime() format string.

localtime

The time at which the filter is running, expressed in the local time zone. It can accept an argument: a strftime() format string.

metadata

Frame metadata. Takes one or two arguments. The first argument is mandatory and specifies the metadata key. The second argument is optional and specifies a default value, used when the metadata key is not found or empty. Available metadata can be identified by inspecting entries starting with TAG included within each frame section printed by running ffprobe -show_frames. String metadata generated in filters leading to the drawtext filter are also available.

n, frame_num

The frame number, starting from 0.

pict_type

A one character description of the current picture type.

pts

The timestamp of the current frame. It can take up to three arguments. The first argument is the format of the timestamp; it defaults to flt for seconds as a decimal number with microsecond accuracy; hms stands for a formatted [-]HH:MM:SS.mmm timestamp with millisecond accuracy. gmtime stands for the timestamp of the frame formatted as UTC time; localtime stands for the timestamp of the frame formatted as local time zone time. The second argument is an offset added to the timestamp. If the format is set to hms, a third argument 24HH may be supplied to present the hour part of the formatted timestamp in 24h format (00-23). If the format is set to localtime or gmtime, a third argument may be supplied: a strftime() format string. By default, YYYY-MM-DD HH:MM:SS format will be used.

Commands

This filter supports altering parameters via commands:

reinit

Alter existing filter parameters. Syntax for the argument is the same as for filter invocation, e.g. fontsize=56:fontcolor=green:text=Hello World Full filter invocation with sendcmd would look like this: sendcmd=c=56.0 drawtext reinit fontsize=56\:fontcolor=green\:text=Hello\\ World

If the entire argument can’t be parsed or applied as valid values then the filter will continue with its existing parameters.

Examples

• Draw Test Text with font FreeSerif, using the default values for the optional parameters. drawtext=“fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text=Test Text”
• Draw ’Test Text’ with font FreeSerif of size 24 at position x=100 and y=50 (counting from the top-left corner of the screen), text is yellow with a red box around it. Both the text and the box have an opacity of 20%. drawtext=“fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text=Test Text:\ x=100: y=50: fontsize=24: fontcolor=yellow@0.2: box=1: boxcolor=red@0.2” Note that the double quotes are not necessary if spaces are not used within the parameter list.
• Show the text at the center of the video frame: drawtext=“fontsize=30:fontfile=FreeSerif.ttf:text=hello world:x=(w-text_w)/2:y=(h-text_h)/2”
• Show the text at a random position, switching to a new position every 30 seconds: drawtext=“fontsize=30:fontfile=FreeSerif.ttf:text=hello world:x=if(eq(mod(t\,30)\,0)\,rand(0\,(w-text_w))\,x):y=if(eq(mod(t\,30)\,0)\,rand(0\,(h-text_h))\,y)”
• Show a text line sliding from right to left in the last row of the video frame. The file LONG_LINE is assumed to contain a single line with no newlines. drawtext=“fontsize=15:fontfile=FreeSerif.ttf:text=LONG_LINE:y=h-line_h:x=-50*t”
• Show the content of file CREDITS off the bottom of the frame and scroll up. drawtext=“fontsize=20:fontfile=FreeSerif.ttf:textfile=CREDITS:y=h-20*t”
• Draw a single green letter g, at the center of the input video. The glyph baseline is placed at half screen height. drawtext=“fontsize=60:fontfile=FreeSerif.ttf:fontcolor=green:text=g:x=(w-max_glyph_w)/2:y=h/2-ascent”
• Show text for 1 second every 3 seconds: drawtext=“fontfile=FreeSerif.ttf:fontcolor=white:x=100:y=x/dar:enable=lt(mod(t\,3)\,1):text=blink”
• Use fontconfig to set the font. Note that the colons need to be escaped. drawtext=fontfile=Linux Libertine O-40\:style=Semibold:text=FFmpeg
• Draw Test Text with font size dependent on height of the video. drawtext=“text=Test Text: fontsize=h/30: x=(w-text_w)/2: y=(h-text_h*2)”
• Print the date of a real-time encoding (see strftime (3)): drawtext=fontfile=FreeSans.ttf:text=%{localtime\:%a %b %d %Y}

• Show text fading in and out (appearing/disappearing): #!/bin/sh DS=1.0

  “color,drawtext=text=TEST:fontsize=50:fontfile=FreeSerif.ttf:fontcolor_expr=ff0000%{eif\\\\: clip(255*(1*between(t\\, $DS + $FID\\, $DE - $FOD) + ((t - $DS)/$FID)*between(t\\, $DS\\, $DS + $FID) + (-(t - $DE)/$FOD)*between(t\\, $DE - $FOD\\, $DE) )\\, 0\\, 255) \\\\: x\\\\: 2 }”

• Horizontally align multiple separate texts. Note that max_glyph_a and the fontsize value are included in the y offset. drawtext=fontfile=FreeSans.ttf:text=DOG:fontsize=24:x=10:y=20+24-max_glyph_a, drawtext=fontfile=FreeSans.ttf:text=cow:fontsize=24:x=80:y=20+24-max_glyph_a
• Plot special lavf.image2dec.source_basename metadata onto each frame if such metadata exists. Otherwise, plot the string NA. Note that image2 demuxer must have option -export_path_metadata 1 for the special metadata fields to be available for filters. drawtext=“fontsize=20:fontcolor=white:fontfile=FreeSans.ttf:text=%{metadata\:lavf.image2dec.source_basename\:NA}:x=10:y=10”

For more information about libfreetype, check: <*http://www.freetype.org/*>.

For more information about fontconfig, check: <*http://freedesktop.org/software/fontconfig/fontconfig-user.html*>.

For more information about libfribidi, check: <*http://fribidi.org/*>.

Detect and draw edges. The filter uses the Canny Edge Detection algorithm.

The filter accepts the following options:

low

high

Set low and high threshold values used by the Canny thresholding algorithm. The high threshold selects the strong edge pixels, which are then connected through 8-connectivity with the weak edge pixels selected by the low threshold. low and high threshold values must be chosen in the range [0,1], and low should be lesser or equal to high. Default value for low is 20/255, and default value for high is 50/255.

mode

Define the drawing mode.

wires

Draw white/gray wires on black background.

colormix

Mix the colors to create a paint/cartoon effect.

canny

Apply Canny edge detector on all selected planes.

Default value is wires.

planes

Select planes for filtering. By default all available planes are filtered.

Examples

• Standard edge detection with custom values for the hysteresis thresholding: edgedetect=low=0.1:high=0.4
• Painting effect without thresholding: edgedetect=mode=colormix:high=0

Apply a posterize effect using the ELBG (Enhanced LBG) algorithm.

For each input image, the filter will compute the optimal mapping from the input to the output given the codebook length, that is the number of distinct output colors.

This filter accepts the following options.

codebook_length, l

Set codebook length. The value must be a positive integer, and represents the number of distinct output colors. Default value is 256.

nb_steps, n

Set the maximum number of iterations to apply for computing the optimal mapping. The higher the value the better the result and the higher the computation time. Default value is 1.

seed, s

Set a random seed, must be an integer included between 0 and UINT32_MAX. If not specified, or if explicitly set to -1, the filter will try to use a good random seed on a best effort basis.

pal8

Set pal8 output pixel format. This option does not work with codebook length greater than 256.

Measure graylevel entropy in histogram of color channels of video frames.

It accepts the following parameters:

mode

Can be either normal or diff. Default is normal. diff mode measures entropy of histogram delta values, absolute differences between neighbour histogram values.

Apply the EPX magnification filter which is designed for pixel art.

It accepts the following option:

1. Set the scaling dimension: 2 for 2xEPX, 3 for 3xEPX. Default is 3.

Set brightness, contrast, saturation and approximate gamma adjustment.

The filter accepts the following options:

contrast

Set the contrast expression. The value must be a float value in range -1000.0 to 1000.0. The default value is 1.

brightness

Set the brightness expression. The value must be a float value in range -1.0 to 1.0. The default value is 0.

saturation

Set the saturation expression. The value must be a float in range 0.0 to 3.0. The default value is 1.

gamma

Set the gamma expression. The value must be a float in range 0.1 to 10.0. The default value is 1.

gamma_r

Set the gamma expression for red. The value must be a float in range 0.1 to 10.0. The default value is 1.

gamma_g

Set the gamma expression for green. The value must be a float in range 0.1 to 10.0. The default value is 1.

gamma_b

Set the gamma expression for blue. The value must be a float in range 0.1 to 10.0. The default value is 1.

gamma_weight

Set the gamma weight expression. It can be used to reduce the effect of a high gamma value on bright image areas, e.g. keep them from getting overamplified and just plain white. The value must be a float in range 0.0 to 1.0. A value of 0.0 turns the gamma correction all the way down while 1.0 leaves it at its full strength. Default is 1.

eval

Set when the expressions for brightness, contrast, saturation and gamma expressions are evaluated. It accepts the following values:

init

only evaluate expressions once during the filter initialization or when a command is processed

frame

evaluate expressions for each incoming frame

Default value is init.