Skip to main content
The Voice tab controls how an agent sounds. It configures text-to-speech (TTS) while the conversation language and speech-to-text behavior remain connected to the language selection in Agent Core.

​
Conversation language

The Voice tab shows the agent’s current Conversation Language as a reference. Change it in Agent Core, where you can also edit the automatic language rules. The current language modes are:
  • Hindi / English (Auto) for Hindi, English, and natural Hinglish switching.
  • Indian Multilingual (Auto) to follow the caller’s latest supported Indian language.
  • Fixed modes for Hindi, Indian English, US English, British English, Bengali, Gujarati, Kannada, Malayalam, Marathi, Odia, Punjabi, Tamil, Telugu, Urdu, and Assamese.
The language mode selects the speech recognizer profile and the language passed to TTS. A voice can still be selected independently, but Studio may show a compatibility warning when the voice is not a good match for the conversation language.

​
Choose a TTS provider and model

The current Studio catalog includes: Select a provider before choosing a voice. If a provider has multiple models, choose the model from the Model selector and then select a compatible voice. For live conversations, choose ElevenLabs Turbo v2.5; Flash v2.5 and Multilingual v2 are available in the voice catalog but are not currently available for live use.

​
Choose a voice

Use the All, Female, or Male filters, and use the language filter when the provider offers voices in more than one language. Sarvam’s Bulbul v3 voices are marked as multilingual; their output language follows the conversation language mode. Select Save Changes to keep the provider, model, and voice on the agent. For a phone agent, choose a voice that is clear at telephone quality, not only pleasant through a laptop speaker.

​
Speaking speed

Providers that support speed show a Speaking Speed control from 0.5x to 2.0x. Start at 1.0x, then adjust after listening to a phone-quality preview.

​
Emotion

Cartesia Sonic 3 exposes an Emotion selector. It includes options such as Neutral, Curious, Excited, Calm, Confident, Apologetic, and Frustrated. Emotion is provider- and model-specific; the control is not shown for providers that do not support it.

​
Preview the voice

In Preview Voice, select Play to synthesize the current greeting. If the agent has a greeting, the preview uses it; otherwise Vocily uses a sample greeting. Choose a quality before listening:
  • Browser (HD) previews at browser-oriented quality.
  • Phone (16kHz) is a closer check of how the selected voice will translate to a call.
Previewing does not save the agent. Select Save Changes after you are happy with the provider, model, voice, speed, and emotion.

​
Voice quality checklist

Test the greeting and at least one longer response in both quality modes. Check that:
  • The first sentence is easy to understand without repeating it.
  • Names, numbers, email addresses, URLs, and IDs are pronounced acceptably.
  • The voice remains intelligible at the chosen speed.
  • The selected voice fits the language mode used by the caller.
  • The agent’s Runtime interruption and response settings do not make the conversation feel rushed or difficult to interrupt.
The voice pipeline uses the selected STT, LLM, and TTS configuration together. If the agent is not recognizing a language correctly, adjust the conversation language in Agent Core before changing the TTS voice.