Add stereo glut engine

[sixolet]

This is an alternate engine, much like original Glut but stereo.

I left it as a separate engine because it does probably use twice the CPU, so people who don't need stereo don't have to have it.

[artfwo]

Hey, thanks for the PR!

Is it possible to load and play mono samples with StereoGlut? To merge the change, I would prefer having the possibility of both playing mono and stereo samples with the same engine.

Maybe with an extra argument in the synth (nchannels) that's being set at sample loading time the engine could be modified to play both stereo and mono samples without having to switch, wdyt?

[sixolet]

Double-check me, but I am pretty sure I coded it to load a mono sample into both channels in StereoGlut, so you can still play mono samples. The architecture of the engine is each channel as a persistent synth, though, so the price you pay is that in StereoGlut you’re still always using the cpu for two channels of playback.

…

On Sun, Dec 4, 2022 at 8:59 AM Artem Popov ***@***.***> wrote: Hey, thanks for the PR! Is it possible to load and play mono samples with StereoGlut? To merge the change, I would prefer having the possibility of both playing mono and stereo samples with the same engine. Maybe with an extra argument in the synth (nchannels) that's being set at sample loading time the engine could be modified to play both stereo and mono samples without having to switch, wdyt? — Reply to this email directly, view it on GitHub <#1 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAMOZYPAQAZ3HTQP6VIH2Y3WLTEYJANCNFSM6AAAAAARMT5KIQ> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[artfwo]

Is the stereo version able to play all the voices together without hitting the CPU limit? If yes, I don't mind converting mono samples to stereo in the original engine.

Alternatively, the read command could be modified to rebuild synths dynamically (check Mass producing SynthDefs tutorial for an example) or a conditional UGen could be used to choose the stereo signal path during synth runtime.

[sixolet]

Yea, it can, with an asterisk. The limitation is the amount of grain overlap-stereo can support less overall grain overlap.

…

On Sun, Dec 4, 2022 at 9:45 AM Artem Popov ***@***.***> wrote: Is the stereo version able to play all the voices together without hitting the CPU limit? If yes, I don't mind converting mono samples to stereo in the original engine. Alternatively, the read command could be modified to rebuild synths dynamically (check Mass producing SynthDefs <http://madskjeldgaard.dk/posts/2019-08-19-supercollider-tutorial-mass-producing-synthdefs/#first-technique-multichannel-enumeration> tutorial for an example) or a conditional UGen <https://doc.sccode.org/Reference/if.html> could be used to choose the stereo signal path during synth runtime. — Reply to this email directly, view it on GitHub <#1 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAMOZYMWVX2HINMYP4YNWATWLTKEHANCNFSM6AAAAAARMT5KIQ> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[sixolet]

The synths are structurally different enough that it’s probably easier just to have them both written out than mass produce them, I think. The conditional ugen might be an option, actually, if it’s upstream of the grain generators so they don’t generate grains on the path they’re not on. I’ll investigate that if you like, that sounds like an interesting way to resolve this.

…

On Sun, Dec 4, 2022 at 9:49 AM Naomi Seyfer ***@***.***> wrote: Yea, it can, with an asterisk. The limitation is the amount of grain overlap-stereo can support less overall grain overlap. On Sun, Dec 4, 2022 at 9:45 AM Artem Popov ***@***.***> wrote: > Is the stereo version able to play all the voices together without > hitting the CPU limit? If yes, I don't mind converting mono samples to > stereo in the original engine. > > Alternatively, the read command could be modified to rebuild synths > dynamically (check Mass producing SynthDefs > <http://madskjeldgaard.dk/posts/2019-08-19-supercollider-tutorial-mass-producing-synthdefs/#first-technique-multichannel-enumeration> > tutorial for an example) or a conditional UGen > <https://doc.sccode.org/Reference/if.html> could be used to choose the > stereo signal path during synth runtime. > > — > Reply to this email directly, view it on GitHub > <#1 (comment)>, or > unsubscribe > <https://github.com/notifications/unsubscribe-auth/AAMOZYMWVX2HINMYP4YNWATWLTKEHANCNFSM6AAAAAARMT5KIQ> > . > You are receiving this because you authored the thread.Message ID: > ***@***.***> >

[artfwo]

Yeah, either way is fine with me. The stereo synths built next to mono versions might be even a cleaner solution. So please go ahead with whichever way you prefer if you'd like to resolve the issue in this repo 👍