Voices (text to speech)

Choose Cartesia, ElevenLabs or OpenAI as your agent's voice, pick a voice, and add one by its ID.

What you will achieve

You will give the agent a voice that fits your business and your callers' language.

Open the agent, Assistant tab, and click the Voice card ("How the agent sounds when it speaks").
In Speaking, choose a provider:
- Cartesia: fastest and most natural.
- ElevenLabs: natural and expressive.
- OpenAI: good quality, slower to start speaking.
In Voice, choose a voice. Each shows a name and a short description such as "Female, friendly supporter". For Cartesia and ElevenLabs the voices offered match the agent's language.
Press Save, and use Talk to listen.

For Cartesia and ElevenLabs, choose Another voice id… and paste the ID.

Cartesia voice id: copy it from play.cartesia.ai.
ElevenLabs voice id: copy it from ElevenLabs' Voice Library. A voice from the library must first be added to My Voices in your ElevenLabs account, or it cannot be used.

Switching provider picks that provider's first voice, because voice IDs only mean something to the provider they came from.
OpenAI voices (Alloy, Echo, Fable, Onyx, Nova, Shimmer) are tuned for English; other languages carry an accent.
For languages other than English, Cartesia uses its newer Sonic 3.6 model; for Urdu, ElevenLabs uses its v4 Turbo model.
Each provider shows its price per minute next to its name.

Use Talk and listen to a full reply, not just the opening line. Check names, numbers and the languages you expect.

The agent speaks in the chosen voice with natural pacing, and the Voice card shows its name and provider.

Something here does not match what you see in the dashboard? Let us know.