-
Restrain buffers piping to node-speaker, bind to RT. Do not generate more than needed.
- node-speaker restrains pressure, but with a 3s buffer - it should be able to be regulated.
-
replace node-speaker with good implementation. It fails on estimating big buffers
-
Decrease buffer, significantly. Make gain work realtime
-
Flash fallback
-
Create scriptProcessorNode mode. As an alternative.
-
For old browsers generate sound like t='data:audio/wav;base64,UklGRl9vT19XQVZFZm10IBAAAAABAAEAQB8AAEAfAAABAAgA',new Audio(t+btoa(t+S)).play()
-
Detect audioBufferSize based on some performance measure, to avoid GC glitches
-
Test in Firefox, Opera, Safari, iOS Safari, IE, others.
-
Test variety of channels
-
Test different sample rates
-
CLI