> ## 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 models

> Speech generation and transcription providers.

Speech routes use the same paid inference pipeline as text and media routes. Text-to-speech can return binary audio instead of JSON.

Provider families with speech or transcription adapter paths include:

| Provider or family | Routes                                     |
| ------------------ | ------------------------------------------ |
| OpenAI             | Speech and transcription.                  |
| Google and Vertex  | Speech generation and transcription paths. |
| Alibaba            | Speech and transcription.                  |
| Cloudflare         | Speech and transcription.                  |
| Deepgram           | Speech generation and transcription.       |
| ElevenLabs         | Speech generation.                         |
| Cartesia           | Speech generation.                         |

Check the catalog row before selecting a model. Voice names, output formats, and language support are provider-specific.

## Related

* [Speech](/inference/modalities/speech)
* [Metering](/inference/metering)