Skip to main content

​
Text-to-speech models

​
v0
default:"default"
The current Chariot TTS model. Produces natural English and Indian-language speech. This is the default; you can also set it explicitly with "model_type": "v0" in the request body.
Modelmodel_typeLanguagesStatus
Chariot TTS v0v0English (en), Hindi (hi)Available
If you omit model_type, requests use v0.

​
Languages

Chariot voices cover English and Hindi today, across American, British, and Indian accents. The language and accent are properties of the voice, not a request parameter, pick a voice in the language you want. Browse and filter the catalog on the Voices page or with GET /v1/voices.

​
Limits

LimitValueApplies to
Max input length500 characters per requestPOST /v1/tts, POST /v1/tts/stream
Output sample rate44.1 kHz, monoAll TTS
Sync output formatWAV (audio/wav)POST /v1/tts
Stream output formatRaw PCM, 16-bit LE (audio/L16)POST /v1/tts/stream
For longer scripts, split the text into chunks of 500 characters or fewer and synthesize each, or use the WebSocket API to feed text incrementally over one connection.
Coming soon: speech-to-text (ASR). It is not yet part of the public API, this page will list ASR models when it ships.