Volume Normalization and Key Binding?

[coolmagnet]

mpv version and platform

mpv-x86_64-20161120 (windows)

The great thing about the VLC player is its adjustable volume normalization feature. VLC appears to be able to even out loud noise, musical score, and spoken dialogue. I've tried the SMPLAYER (with MPV) volume normalization and it does not appear to even out the sound. Sometimes the musical score is louder than the spoken dialogue. So now I'm using MPV without SMPLAYER and everything is working great except I don't know how to turn ON volume normalization and how to key bind it (input.conf)? I'm hoping that MPV would be somewhat equivalent to VLC volume normalization.

Thanks

[garoto]

mpv uses standard ffmpeg audio/video filters and their corresponding syntax detailed here:

https://ffmpeg.org/ffmpeg-filters.html

For audio normalization, I guess the consesus while reading this issue tracker and elsewhere, is that dynaudnorm was the one to use. There's also loudnorm, but I'm not sure if it's even the same thing (don't know much about audio shenanigans).

for keybinding, I have F1 af toggle "lavfi=[dynaudnorm=f=75:g=25:n=0:p=0.58],volume=softclip=yes" in my input.conf that I use for 5.1 content mostly (films) with my headphones and onboard audio.

[coolmagnet]

Thanks garoto.

I also found the following to work:

--af=drc=1:0.25

So I now have keybinding as follows:

F1 af toggle "drc=1:0.25"
F2 af toggle "lavfi=[dynaudnorm=f=75:g=25:n=0:p=0.58],volume=softclip=yes"

They both sound similar but the keybinding allows me to change when needed.

[kevmitch]

Note that last I checked, dynaudionorm didn't play well with video-sync=display-resample due to the large lookahead audio buffer involved. drc, on the other hand is extremely primitive and is almost guaranteed to clip the audio signal. Haven't looked too closely at loudnorm.

[ghost]

That's because speed changes aren't tracked well with the large buffer.

In theory you could add dynaudionorm to --lavfi-complex, but this would "break" track selection currently. It should work with DS, because the "normal" filter chain, which also resamples the audio according to playback speed, won't have the large buffer.

[NightMachinery]

@coolmagnet commented on Jan 6, 2017, 1:48 PM GMT+3:30:

Thanks garoto.

I also found the following to work:

--af=drc=1:0.25

So I now have keybinding as follows:

F1 af toggle "drc=1:0.25"
F2 af toggle "lavfi=[dynaudnorm=f=75:g=25:n=0:p=0.58],volume=softclip=yes"

They both sound similar but the keybinding allows me to change when needed.

What's the CLI option for the second one? I tried --af-add='[dynaudnorm=f=75:g=25:n=0:p=0.58],volume=softclip=yes', but it gives a parsing error.

Update: --af-add='dynaudnorm=g=5:f=250:r=0.9:p=0.5' works.