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

# Speech

> Text-to-speech and speech-to-text routes.

Speech routes handle audio generation and transcription.

| Method | Path                                | Description                                 |
| ------ | ----------------------------------- | ------------------------------------------- |
| `POST` | `/v1/audio/speech`                  | Text to speech. Returns audio bytes.        |
| `POST` | `/v1/audio/transcriptions`          | Speech to text. Returns JSON or plain text. |
| `POST` | `/external/v1/audio/speech`         | External text to speech.                    |
| `POST` | `/external/v1/audio/transcriptions` | External transcription.                     |

## Text to speech

```json theme={null}
{
  "model": "tts-1",
  "input": "Compose settles paid model calls over HTTP.",
  "voice": "alloy",
  "response_format": "mp3",
  "speed": 1
}
```

The response body is audio. The `Content-Type` comes from provider output, usually `audio/mpeg` or the requested format.

## Transcription

```json theme={null}
{
  "model": "whisper-1",
  "file": "<base64 audio>",
  "language": "en",
  "response_format": "json"
}
```

Set `response_format` to `text` when you want a plain text body.

## Related

* [Metering](/inference/metering)
* [Model families](/inference/families/speech-models)