> ## Documentation Index
> Fetch the complete documentation index at: https://docs.outspeed.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Text-to-Speech (HTTP)

> Generate WAV audio from text using the HTTP TTS endpoint.

## Overview

Generate audio from text using our TTS endpoints. We support both single-voice and multi-voice dialogue generation.

* **Single Voice TTS**: Convert text to speech with one voice
* **Dialogue Generation**: Mix narrator and character voices in the same audio file

## Single Voice TTS

POST `https://api.outspeed.com/v1/tts/`

### Request Body

```json theme={null}
{
  "model": "outspeed-tts-v2",
  "voice": "clark",
  "text": "Hello, world!",
  "stream": false
}
```

* **model**: TTS model to use. Use `outspeed-tts-v2` (`outspeed-tts-v1` is deprecated)
* **voice**: the voice identifier. Find all available voices and their models at [TTS Playground](https://dashboard.outspeed.com/tts)
* **text**: the text to synthesize
* **stream**: set to `true` to stream audio chunks; `false` returns the full WAV

### Response

* Content-Type: `audio/pcm`
* Headers:
  * `X-Sample-Rate`: Sample rate (default: 24000)
  * `X-Channels`: Number of audio channels (default: 1)
  * `X-Bit-Depth`: Bit depth (default: 16)
* Body: Raw PCM audio bytes (little-endian int16), 24kHz, mono. No WAV header is included. Wrap with a WAV header or convert with a tool like ffmpeg.

<Info>
  Authenticate with <code>Authorization: Bearer \<YOUR\_OUTSPEED\_API\_KEY></code>.
</Info>

## Examples (non-streaming)

<CodeGroup>
  ```bash curl theme={null}
  curl \
    -X POST \
    'https://api.outspeed.com/v1/tts/' \
    -H 'Authorization: Bearer YOUR_OUTSPEED_API_KEY' \
    -H 'Content-Type: application/json' \
    -d '{"model":"outspeed-tts-v2","voice":"clark","text":"Hello, world!","stream":false}' \
    --output tts.pcm
  ```

  ```bash javascript theme={null}
  // Node.js 18+

  //Use a WAV helper library to write a proper WAV file from PCM:
  //npm install wav

  import wav from "wav";

  const res = await fetch("https://api.outspeed.com/v1/tts/", {
    method: "POST",
    headers: {
      Authorization: "Bearer YOUR_OUTSPEED_API_KEY",
      "Content-Type": "application/json",
    },
    body: JSON.stringify({ model: "outspeed-tts-v2", voice: "clark", text: "Hello, world!", stream: false }),
  });

  if (!res.ok) throw new Error(`TTS failed: ${res.status}`);
  const pcm = Buffer.from(await res.arrayBuffer());
  const writer = new wav.FileWriter("tts.wav", { channels: 1, sampleRate: 24000, bitDepth: 16 });
  writer.write(pcm);
  writer.end();
  ```
</CodeGroup>

## Examples (streaming)

<CodeGroup>
  ```bash curl theme={null}
  curl \
    -X POST \
    'https://api.outspeed.com/v1/tts/' \
    -H 'Authorization: Bearer YOUR_OUTSPEED_API_KEY' \
    -H 'Content-Type: application/json' \
    -d '{"model":"outspeed-tts-v2","voice":"clark","text":"Hello, world!","stream":true}' \
    --no-buffer \
    --output tts.pcm
  ```

  ```bash javascript theme={null}
  // Node.js 18+ (streaming)

  //Use a WAV helper library to write a proper WAV file from PCM:
  //npm install wav

  import { Readable } from "node:stream";
  import wav from "wav";

  const res = await fetch("https://api.outspeed.com/v1/tts/", {
    method: "POST",
    headers: {
      Authorization: "Bearer YOUR_OUTSPEED_API_KEY",
      "Content-Type": "application/json",
    },
    body: JSON.stringify({ model: "outspeed-tts-v2", voice: "clark", text: "Hello, world!", stream: true }),
  });

  if (!res.ok || !res.body) throw new Error(`TTS failed: ${res.status}`);
  const nodeReadable = Readable.fromWeb(res.body);
  const writer = new wav.FileWriter("tts.wav", { channels: 1, sampleRate: 24000, bitDepth: 16 });
  await new Promise((resolve, reject) => {
    nodeReadable.pipe(writer);
    writer.on("finish", resolve);
    writer.on("error", reject);
    nodeReadable.on("error", reject);
  });
  ```
</CodeGroup>

## Dialogue Generation

Generate audio with multiple voices (narrator + character) using the `outspeed-tts-v2` model.

<Note>Try it visually at the [Dialogue Playground](https://dashboard.outspeed.com/tts/dialogue)</Note>

### Endpoint

POST `https://api.outspeed.com/v1/tts/dialogue`

### Request Body

```json theme={null}
{
  "model": "outspeed-tts-v2",
  "text": "*The old library stood silent.* I pushed open the heavy door. *Inside, dust particles danced in my flashlight beam.*",
  "speaker_voice": "9c5c73f4-1cb7-46cf-91d8-24c80b6288f0",
  "narrator_voice": "a42c84b2-0e6b-4c9f-a8e7-3f5d1c2e8a9b",
  "narrator_delimiter": "*"
}
```

### Parameters

| Parameter            | Type   | Required | Description                                         |
| -------------------- | ------ | -------- | --------------------------------------------------- |
| `model`              | string | Yes      | Must be `outspeed-tts-v2`                           |
| `text`               | string | Yes      | Dialogue text with delimiters for narrator parts    |
| `speaker_voice`      | string | Yes      | Voice ID for character dialogue                     |
| `narrator_voice`     | string | No       | Voice ID for narrator parts (omit to skip narrator) |
| `narrator_delimiter` | string | No       | Delimiter: `*`, `(`, `[`, or `{` (default: `*`)     |

### Delimiter Usage

* **Text in delimiters** = narrator voice
* **Text outside delimiters** = character voice

Supported delimiters:

* `*` → `*narrator text*`
* `(` → `(narrator text)`
* `[` → `[narrator text]`
* `{` → `{narrator text}`

### Examples

<CodeGroup>
  ```bash Audiobook Style theme={null}
  curl -X POST https://api.outspeed.com/v1/tts/dialogue \
    -H "Authorization: Bearer YOUR_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "model": "outspeed-tts-v2",
      "text": "*The old library stood silent.* I pushed open the heavy door. *Inside, dust particles danced in my flashlight beam.*",
      "speaker_voice": "<voice-id-1>",
      "narrator_voice": "<voice-id-2>,
      "narrator_delimiter": "*"
    }' \
    --output dialogue.wav
  ```
</CodeGroup>

### Voice Selection

Both voices require **voice IDs** (not names). Get voice IDs from:

* [TTS Playground](https://dashboard.outspeed.com/tts) - Browse and copy existing voice IDs
* [Voice Upload](/features/voice-cloning) - Upload custom voices or create [voice clones](/features/voice-cloning)

### Response

Same as single voice TTS:

* Content-Type: `audio/wav`
* Headers: `X-Sample-Rate`, `X-Channels`, `X-Bit-Depth`
* Body: Raw PCM audio bytes