Cartesia
Cartesia offers a wide selection of fully multilingual voices with very low latency. Create a Cartesia account to browse and test voices in the Cartesia Playground.
Models
Cartesia provides multiple generations of its Sonic TTS model:
All Cartesia voices can be used with any model.
Learn about Cartesia’s latest Sonic 3 model.
Documentation for older Sonic model versions.
Voices
Copy the voice ID from the table below:
For more information, refer to Cartesia’s guide to Choosing a Voice.
Usage
Cartesia voice IDs conform to the following format:
Parameters:
voice_id(required): The UUID voice identifier from the Voices tablemodel(optional): One of the Sonic models listed above (default:sonic-3)
Examples:
Languages
Cartesia voices are fully multilingual when used with sonic-multilingual, sonic-2, sonic-3, or sonic-3 models.
The multilingual models automatically adapt to the input text language.
Supported languages include: English, Spanish, French, German, Italian, Portuguese, Dutch, Polish, Russian, Chinese, Japanese, Korean, Hindi, Turkish, Swedish, and many more.
For the complete list, refer to Cartesia’s Sonic 3 language support and Sonic 2 language support references.
Examples
See how to use Cartesia voices on the SignalWire platform.
SWML
RELAY Realtime SDK
Call Flow Builder
Use the
languages
SWML method to set one or more voices for an AI agent.
Alternatively, use the say_voice parameter
of the play
SWML method to select a voice for basic TTS.