01 / DECIDE
Architecture decision
Use it for low-latency speech interaction where file-based audio would feel discontinuous.
Transport, ephemeral access, turn detection, interruption, tools, reconnect, and playback state jointly define the experience. Map input, output, state, tools, and side effects as one observable path, then name the owner of design voice as a live session.
02 / OPERATE
Three-step field procedure
- 01
Lock the contract
Write down the caller, data classification, success condition, timeout, cancellation, and ownership. Use it for low-latency speech interaction where file-based audio would feel discontinuous.
- 02
Run one bounded path
Build one end-to-end path with correlation, typed state, and reversible failure handling. Transport, ephemeral access, turn detection, interruption, tools, reconnect, and playback state jointly define the experience.
- 03
Verify the signal
Turn acceptance into a repeatable fixture, contract test, or browser test. Tests cover jitter, double-talk, reconnect, late tools, device changes, and clean shutdown.
03 / BOUND
Operational boundary
Never expose long-lived keys to clients; interruption must cancel stale audio and tool work.
04 / PROVE
Acceptance evidence
Tests cover jitter, double-talk, reconnect, late tools, device changes, and clean shutdown.
SOURCE / HTTP
Reproducible source probe
curl -fsSI 'https://docs.x.ai/developers/model-capabilities/audio/voice' | sed -n '1,5p'