AGN-672 Voice Benchmark
Internal synthetic testing only. Audio, transcripts, conversation history and tokens are not persisted by this application.
Audio input
No recording in progress.
No audio selected.
Result
- Latency
- Transcript
- Response
- Response audio
Streaming conversation
Speak naturally after the state becomes listening. Speaking over agent audio interrupts playback and returns the session to listening.
- State
- ended
- Diagnostic correlation
- Not issued
- End of turn → first audible response
- Not observed
- End of turn → completed response
- Not observed
- Turn/session failures
- 0
- Interruptions
- 0