[FFmpeg-user] amix: Hard limit for number of inputs?

Paul B Mahol onemda at gmail.com
Tue Jul 16 18:45:55 EEST 2019

On 7/16/19, Reino Wijnsma <rwijnsma at xs4all.nl> wrote:
> On 16-7-2019 15:36, Paul B Mahol <onemda at gmail.com> wrote:
>> On 7/16/19, Reino Wijnsma <rwijnsma at xs4all.nl> wrote:
>>> About 7 years ago I've used Audacity to assemble the soundtrack of the
>>> videogame No One Lives Forever 2. See
>>> https://www.youtube.com/watch?v=4Y3aKcQ0HK4 for example.
>> Why you use adelay?
>> Perhaps you want acrossfade filter instead, or even better concat filter.
> I can't use these filters. Have a look at the Youtube video please.
> I wish it was a simple matter of just concatenating all segments, but it
> isn't, because the segments don't align perfectly.
> Each segments has to start at a very specific time.
> Then it's a matter of mixing everything at full volume(!), so there's no
> fading involved at all.

acrossfade can just join it with nofade curve, which basically disables fading.

> Do you happen to know why ffmpeg reports a completely wrong duration?
> Related question/suggestion:
> It's really cumbersome to always having to specify the delay for all(!)
> channels: [1]adelay=158792S|158792S[E2];
> How about a default behaviour where specifying an amount of
> milliseconds/samples once(!) applies to all channels: [1]adelay=158792S[E2];
> And if you want a specific delay for another channel, then you can do so as
> is already possible.
> This would make my commandline a lot shorter!

The thing with amix is that by reducing volume by mixing and adding it
back with volume,
you basically loose quality. It is more obvious with float sample
format then double one.

Option could be added to amix filter to just sum samples as is.

Also what ffmpeg version you are using?

More information about the ffmpeg-user mailing list