rason-yang

Do you send the model mixed audio or s separate track for everyone ? When people talk at the same time, how to speaker diarization ？

This matches how i think about realtime voice: keep call control deterministic, and give model steps clear inputs and outputs. How would you handle retries when a step fails ?

I build low-latency voice gateways . Does your local API expose interim transcripts and timestamps while audio is arriving , or only the final text after a dictation ends ?

i work on real-time voice AI systems. how does Tachyon handle cancellation and client disconnects for long-running MCP tools over Streamable HTTP?