I've experimented with adding sound to my TRS-80 emulator. It seems like the only workable solution is to use the ScriptProcessorNode callback as the sole source to timing and run the emulator from there. This will still jitter if Javascript is too slow. But if the emulator is too slow, then, there is nothing you can do except report it to the user.
The graphics are done on an ordinary timer which redraws based on the cached video state from the audio callback. JSMESS could probably do the same.
You'll always have a serious problem when there is more than one timer. If I could change the APIs, I'd add a requestAudioVideoFrame() timer callback where you would supply both the video (rendering) and the audio buffer for a given segment of time. That would also give the browser more control over the situation.
The graphics are done on an ordinary timer which redraws based on the cached video state from the audio callback. JSMESS could probably do the same.
You'll always have a serious problem when there is more than one timer. If I could change the APIs, I'd add a requestAudioVideoFrame() timer callback where you would supply both the video (rendering) and the audio buffer for a given segment of time. That would also give the browser more control over the situation.