This should be possible from the browser APIs but it looks like offline audio node will not work, so it's probably not possible to extract audio from video, faster than real-time.
The alternative is to parse files directly, using something like https://github.com/gpac/mp4box.js/ (having looked into browser apis, I think direct parsing is more promising)
Ultimately it would be possible to reimplement aikuma features using a video playback system instead but it raises questions about whether it's possible to build a system that works equally well from video or audio. The most robust approach may be to treat video as optional and simply trigger a player with the same time queues as we trigger our audio playback schemes.
If that's the right approach, it comes back to needing to extract audio from video again.
This should be possible from the browser APIs but it looks like offline audio node will not work, so it's probably not possible to extract audio from video, faster than real-time.
The alternative is to parse files directly, using something like https://github.com/gpac/mp4box.js/ (having looked into browser apis, I think direct parsing is more promising)
Ultimately it would be possible to reimplement aikuma features using a video playback system instead but it raises questions about whether it's possible to build a system that works equally well from video or audio. The most robust approach may be to treat video as optional and simply trigger a player with the same time queues as we trigger our audio playback schemes.
If that's the right approach, it comes back to needing to extract audio from video again.