[whimsicalism]

Makes sense, I think streaming audio->audio inference is a relatively big lift.

[red2awn]

Correct, it's breaks the single prompt, single completion assumption baked into the frameworks. Conceptually it's still prompt/completion but for low latency response you have to do streaming KV cache prefill with a websocket server.

[whimsicalism]

I imagine you have to start decoding many speculative completions in parallel to have true low latency.