Start audio playback

**Rate limit: Per command** · **Burst:** 8 · **Refill:** 4 req/s Choose the tab for your audio source. <Tabs> <Tab title="Audio files"> Play one WAV or MP3 file, or a playlist: ```json { "type": "files", "urls": [ "https://media.example.com/greeting.mp3", "https://media.example.com/menu.wav" ] } ``` The platform plays the URLs in order. Use `offset_ms` only with one file. For private files, use presigned HTTPS URLs that remain valid through playback. [Seek within playback](/api/playback/seek) is available only for files. </Tab> <Tab title="Live TTS"> Start a pushed raw-audio stream: ```json { "type": "stream", "media_format": { "encoding": "pcm_s16le", "sample_rate": 24000 } } ``` The `202 Accepted` response includes a short-lived `upload_url`. Stream the TTS bytes in one HTTP request body, then close the body when the TTS stream ends: ```http PUT {upload_url} Content-Type: application/octet-stream <raw audio bytes> ``` The audio must match `media_format`: signed little-endian PCM16 at 8, 16, 24, or 48 kHz; or G.711 mu-law at 8 kHz. </Tab> <Tab title="Silence"> Keep Session Playback active without an audio source: ```json { "type": "silence" } ``` Silence continues until another playback replaces it, the session ends, or you [stop playback](/api/playback/stop). </Tab> </Tabs> An accepted request returns `202 Accepted` with `operation_uuid` and `status: in_progress`. New playback replaces active playback automatically. See [Playback lifecycle events](/api/playback#lifecycle-events) for completion and failure webhooks.