[opus] Antw: Re: Antw: [EXT] Opus merging streams

[Ulrich Windl]

>>> Andrew Sonzogni <andrew at safehear.fr> schrieb am 06.04.2022 um 09:47 in
Nachricht <f7a34d74-0026-1d71-6823-62454988aa33 at safehear.fr>:
> Thanks for the info !
> 
> Then I have a question. What kind of mixer algorithms can be used to mix 
> 3 different channel together on an embedded system ?
> 
> I've used this one but it's not THAT good: (chan1 + chan2 + chan3) / 3

Obviously if you have an overflow first, then dividing by three won't help, I
guess.

chan1/3 + chan2/3 + chan3/3 ?

> 
> The output signal may peak or be buggy some times.
> 
> For your information, I'm using an ARM M4F with Opus configured like 
> this (40ms, 16kHz, 16 bitrate, 0 compres).

Regards,
Ulrich

> 
> 
> Kind regards,
> 
> Andrew
> 
> Le 06/04/2022 à 08:36, Sampo Syreeni a écrit :
>> On 2022-04-06, Ulrich Windl wrote:
>>
>>> Incidentially I came across a Dolby Atmos demo that had about 118 
>>> channels wirh 24bit audio at 48kHz, all in one huge WAV file yesterday.
>>
>> Is that even a legitimate encoding?! What the fuck.
>>
>>> When I tried to play that (in plain stereo) with audiacity, even my 
>>> fast computer (i7 at 4GHz) had dropouts. So I can imagine that 
>>> decoding a large number of channels and mixing those seems to be a 
>>> bad idea.
>>
>> It is. Which is why my favourite ambisonics exists (sales pitch): it's 
>> a principled and nigh entropically speaking optimum way to fold down a 
>> static central soundfield down to a number of channels. Third order, 
>> so sixteen channels, seems to be upto the task for *any* central 
>> isotropic soundfield at all, and the system yields to static 
>> optimization.
>>
>> I cannot for the life of me understand why Atmos exists. Except for 
>> patent patent law or something like that. If it was used to express a 
>> live gaming or augmented reality setup, with arbitrary auditory 
>> parallax, I could get the point. But that's not what Atmos or even 
>> Dolby AC-4 are about. They just encode a static scene -- in a way 
>> *much* more complicated and heavier on the processor than a "simple" 
>> third degree periphonic ambisonic HOA signal set would be, and in a 
>> manner not amenable to low resource optimizations in surround sound. 
>> The object based encoding simply seems stupid and superfluous.
> -- 
> 
>