---
title: "Streaming text-to-speech synthesis (HTTP)"
method: POST
path: "/stream"
tags: ["textToSpeech"]
---

# Streaming text-to-speech synthesis (HTTP)

`POST /stream`

Stream audio as it's generated. Returns chunked WAV data for progressive playback. The model associated with the voice is selected automatically.

## Headers

- `Authorization` string, required

## Request body

- object
  - `voice_uuid` string, required — Voice UUID to use for synthesis
  - `data` string, required — Text or SSML to synthesize (max 2000 characters)
  - `project_uuid` string — Optional project UUID to store the clip
  - `precision` 'MULAW' | 'PCM_16' | 'PCM_24' | 'PCM_32' — Audio precision
  - `sample_rate` '8000' | '16000' | '22050' | '32000' | '44100' | '48000' — Audio sample rate in Hz
  - `use_hd` boolean — Enable HD synthesis with small latency trade-off
  - `apply_custom_pronunciations` boolean — When true, automatically applies your team's custom pronunciations to matching words in the input text. Defaults to false.

## Response `200`

Streaming audio response (chunked WAV)

## Other responses

- `400` — Bad request

---

[API](https://skmtc.net/resemble/apis/api-reference.md) · [All operations](https://skmtc.net/resemble/apis/api-reference/llms.txt) · [OpenAPI document](https://skmtc-service-staging.skmtc.workers.dev/v1/apis/resemble/api-reference/versions/bce48b20dad7/schema)