[FFmpeg-devel] [PATCH] Add NULL mixer to AVSequencer.

[Sebastian Vater]

Ronald S. Bultje a ?crit :
> Hi,
>
> On Mon, Aug 16, 2010 at 2:22 AM, Sebastian Vater
> <cdgs.basty at googlemail.com> wrote:
>   
>> We should utilize multiple mixers any way. Since we want to target
>> multiple audiences, right?
>>     
>
> Do we want multiple mixers? I'm trying to make the point that we
> don't. If the quality of a S32 mixer is poor, than your mixer isn't
> good enough and we need to improve it before committing it. I've never
> heard of anyone wanting to sacrifice 99% of the sound quality for a 2x
> speedup, so to say.
>   

Well it's more 90% for 20x speedup, the high quality mixer is really way
slower than the low quality mixer. At normal 44,1kHz they don't differ
much (unless you have an $8000 audio equipment and excellent ears).

Or to ask another question, you surely tried mod playback with my latest
patches, would you consider the sound quality bad, esp. bad enough to
not be acceptable? This configuration used by IFF-TCM1 demuxer is really
the lowest quality configuration of it being possible.

> Try to write your "null mixer" as a if (!buf) in the high-quality
> mixer, that should suffice for both cases, saves >1000 lines of code
> and means we don't need a synth mixer registration API and versioning
> symbols anymore. Also, only do this if this is strictly required for
> TCM1. If there's other ways to accomplish the same thing in TCM1,
> don't focus on the null mixer, since then we clearly don't need it.
>   

Well, instead of doing the buf == NULL stuff, I would recommend setting
volume_boost to 0 in that case. This would be a much more safer option
and also wouldn't require any external app checks (in worst case they
could reset boost to the value before).

Also what's about the OPL2/3 stuff, and SID stuff what we might add
later? Or do you mean we simply could not use a multiple mixers
interface unless we really need it? In that case I agree!

-- 

Best regards,
                   :-) Basty/CDGS (-: