Discrete voice audio routing

[javigaralva]

Export channels to wav files could be a nice feature. Could be possible to do that? Currently I try to do this job with RYMCast, but the channels completely panned to the right, generate an empty wav file.

Thanks for this amazing project

[mmontag]

Basically need 2 things in order to do this...

• refactor audio output to a generic interface, which could be either a file writer or WebAudio context. Medium difficulty.
• expose discrete voices (write to an audio buffer for each voice) outside of each player engine. The player engines are all designed for internal mixing, and write to stereo audio buffer only. Very difficulty.

Getting discrete voices in some common interface would have other benefits, like being able to show an oscilloscope for each channel, but it requires modifying the engines. The problem with modifying the engines is that (1) they each do things differently, so the work doesn't scale very well and (2) it makes things (more) brittle since they will diverge from upstream improvements.

I would like to see how Modizer handles this :)

[mmontag]

It looks like Modizer only does this where the engine already supports it.

https://github.com/yoyofr/modizer/blob/master/src/ModizMusicPlayer.mm

see

• m_voicesdataavail=
• m_voice_buff_ana
• //copy voice data for oscillo view

edit: actually Modizer makes deep modification to emulation cores:

https://github.com/yoyofr/modizer/commit/cf117601dd9909b1635b99650df6537f0450b92c#diff-7390462695712dbe8460edcaab01bd6597b9d920f20a15211b0172acc8ceb4df

[mmontag]

Furnace Tracker provides another example of emulation cores being modified to expose discrete channel output:

https://github.com/search?q=repo%3Atildearrow%2Ffurnace+oscilloscope&type=commits&s=committer-date&o=asc

[mmontag]

See also: https://github.com/tildearrow/furnace/discussions/1361 https://github.com/tildearrow/furnace/discussions/1355