As discussed at TPAC 2025, we will accept renderSizeHints in the range of 1 sample to 6 seconds at the context sample rate, and reject all other values. This should be specified in the text.
The reasoning for this range is that we want to support the largest ScriptProcessorNode buffer size (16384) at the lowest sample rate (3000 Hz as sepcified in https://webaudio.github.io/web-audio-api/#sample-rates), to allow existing ScriptProcessorNode applications to be ported to AudioWorklet. This reasoning should be included in a non-normative note in the spec.
As discussed at TPAC 2025, we will accept renderSizeHints in the range of 1 sample to 6 seconds at the context sample rate, and reject all other values. This should be specified in the text.
The reasoning for this range is that we want to support the largest ScriptProcessorNode buffer size (16384) at the lowest sample rate (3000 Hz as sepcified in https://webaudio.github.io/web-audio-api/#sample-rates), to allow existing ScriptProcessorNode applications to be ported to AudioWorklet. This reasoning should be included in a non-normative note in the spec.