Would love to understand how to build audio applications without requiring SharedArrayBuffer

[patrick99e99]

I have been working on a project recently that involves the web audio api, in particular, interacting with emulators of various sound chips.

I followed this guide:
https://emscripten.org/docs/api_reference/wasm_audio_worklets.html?highlight=audio%20worklet

I ended up making multiple calls to emscripten_create_wasm_audio_worklet_processor_async so that I have a worklet running for each sound chip.

What I am trying to do is make it so that my JS code can trigger a sound chip, tell it to play, and then call emscripten_audio_worklet_post_function_v to tell the main thread when it's finished playing. This was all fine and dandy, until I realized compiling with -s AUDIO_WORKLET=1 -sWASM_WORKERS=1 expects there to be a SharedArrayBuffer available...

Ultimately this project is going to be hosted on multiple platforms on servers that I do not control, and requiring them to have SharedArrayBuffer accessible is problematic.

So, I look at projects like snes9x (super nintendo emulator), which has been ported to the browser using emscripten, and that generates real-time audio no problem without the requiring a crossOriginIsolated environment... That makes me think that I am going about this all wrong, and there should be a way for me to do what I want to do without needing SharedArrayBuffer.

I am hoping someone can shed some light on this, and point me in the right direction as far as how I can generate real-time audio, and let my front end JS code know when it's done, without requiring SharedArrayBuffer.

[sbc100]

The AUDIO_WORKLET API in emscripten is fundamentally based on SharedArrayBuffer so if you can't use SharedArrayBuffer then I think your only option it to use a different audio API. Have you looked into using OpenAL , for example: https://emscripten.org/docs/porting/Audio.html