FAQ

​

Audio models

​

What model should I use for text-to-speech?

Use stepaudio-2.5-tts — our flagship contextual TTS model with zero-shot voice cloning and natural-language control over emotion and style through the instruction field and inline () prompts. See Audio Models for details.

​

Where can I find the TTS API parameters?

See Generate audio for the full request schema and examples.

​

What audio formats are supported?

wav, mp3, flac, opus, pcm. Default is mp3.

​

Is there a limit on input length?

Yes. The maximum input length is 1,000 characters per request.