[Ffmpeg-devel] Audio normalization support

[Aaron Williams]

Hi,

I have a project where I need to organize a bunch of videos for playback
through a hardware MPEG player but one problem I have is various videos
people have submitted have different audio levels.  I need to normalize
them and found a GPL tool which can calculate how much the volume needs to
be adjusted (normalize-0.7.7) and was wonering if people would find
something like this useful for ffmpeg?

My idea is to create a new output "codec" which takes the audio samples
and calculates the audio level and saves it to a file.  This could be used
in a second pass to adjust the volume level to some pre-determined level.
This would be analagous to 2-pass video encoding.

At this point, as far as I can tell all I need to do is write a codec and
add support for reading the volume information from the file generated in
the first pass similar to how the -vol parameter works.

Any ideas or suggestions?  I have not written any codecs for ffmpeg
before, but it looks pretty straightforward.  Or perhaps instead of a
codec should I do something else for this?

As far as normalize being GPL, I want to find just the equations used and
re-write a LGPL version for inclusion to ffmpeg, unless the author of
normalize OKs a LGPL version.

-Aaron Williams