[FFmpeg-user] Preserving preserved loudness when downmixing audio from 5.1 AC3 to stereo AAC

[Carl Eugen Hoyos]

Francois Visagie <francois.visagie <at> gmail.com> writes:

> . output from '-ac 2' is noticeably softer than the original
>
> . '-filter:a aformat=channel_layouts=stereo' seems to 
> have the same volume as the original
> 
> . '-filter:a_aresample=ocl=3' is noticeably softer 
> than the original

I tested the following three command lines with random 
5.1 ac3 input:
$ ffmpeg -i input -ac 2 out1.wav
$ ffmpeg -i input -filter:a format=channel_layouts=stereo out2.wav
$ ffmpeg -i input -filter:a aresample=ocl=3 out3.wav

As expected, the three output files are bit-identical.
(Both format and ac 2 auto-insert the aresample filter 
as loglevel debug will tell you.)

The following makes a difference, it does only work 
for certain input codecs though (not aac):
$ ffmpeg -request_channels 2 -i input out4.wav

Carl Eugen