Streaming Audio

Stream realtime speech with sonus-tts using chunked transfer and 16kHz PCM frames.

POST https://api.sagea.space/v1/audio/stream

Authentication

Send Authorization: Bearer $SAGEA_API_KEY on every request. Get a key

Parameters

Send a JSON body with the following fields.

Parameter	Type	Required	Description
`model`	`string`	Yes	Always `sonus-tts` for streaming synthesis.
`input`	`string`	Yes	Text to stream, max 4096 characters per request.
`voice`	`string`	No	One of `nova`, `atlas`, or a cloned `voice_id`.
`emotion`	`string`	No	One of `neutral`, `friendly`, `professional`, `excited`, `calm`, or `warm`.
`speed`	`number`	No	Playback speed from 0.5 to 2.0. Default is `1.0`.
`format`	`string`	No	Frame encoding. Use `pcm16` for 16kHz PCM frames.
`language`	`string`	No	BCP-47 hint such as `ne-NP`, `hi-IN`, `en-US`, or `es-ES`.

Example request

Stream Welcome to SAGEA with a ne-NP greeting in realtime.

curl -N -X POST https://api.sagea.space/v1/audio/stream \
  -H "Authorization: Bearer $SAGEA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "sonus-tts",
    "input": "Welcome to SAGEA. नमस्ते, म तपाईंलाई सहयोग गर्नेछु।",
    "voice": "atlas",
    "emotion": "friendly",
    "speed": 1.0,
    "format": "pcm16"
  }' \
  --output stream.pcm

First audio byte arrives in under 200ms for responsive voice agents.

Response

The API returns chunked 16kHz PCM frames, not a single JSON object.

{
  "note": "Chunked PCM stream. Each frame is 16kHz mono audio.",
  "first_byte_ms": 172,
  "format": "pcm16",
  "sample_rate_hz": 16000,
  "request_id": "stream_atlas77np",
  "model": "sonus-tts",
  "voice": "atlas"
}

Buffer frames client-side and play continuously for realtime output.

Errors

Status	Code	Fix
`400`	`text_too_long`	Split `input` into chunks under 4096 characters.
`401`	`unauthorized`	Re-export a valid `SAGEA_API_KEY` value.
`422`	`voice_not_found`	Use `nova`, `atlas`, or a valid `vox_` voice ID.
`429`	`rate_limit_exceeded`	Back off and retry with exponential delay.
`500`	`internal_error`	Reconnect once, then report request ID.

Limits

Audio streaming allows 50 requests per minute on Starter and 500 requests per minute on Pro.