Currently AudioWorkletProcessor.process() method passes its arguments as `sequence<sequence<Float32Array>>`. The Audio WG discovered that this forces the implementation to allocate memory every processing block causing audio rendering performance degradation over time due to the garbage collection. Per the recent resolution in Audio WG (see the spec link), the WG decided to use FrozenArray<FrozenArray<Float32Array>> instead of compound `sequence` to avoid redundant reallocation.
Specification being incubated in a Community Group
Status in Chromium
Consensus & Standardization
- No signal
- Strongly positive
Search tagsaudio, worklet, wasm,
Last updated on 2022-01-14