TTS Stream

Exercises POST /api/tts/stream — standalone text to speech, outside any conversation session. The response is raw headerless PCM, so this page decodes and schedules it with the Web Audio API rather than handing it to an <audio> element. Sample depth and byte order are fixed by the TTS port (pcm = int16 little-endian); the sample rate varies per deployment, so it is read from X-Audio-Sample-Rate.

Response headers — the playback contract

no request yet

Stream metrics

Time to first byte
Time to first audio (playback starts)
Time to full stream
Chunks received
Bytes received
Audio duration
Faster than realtime
Odd-byte splits handled