Build agentsConfigure your agent

Voice configuration

The Voice tab allows you to customize how your agent sounds and responds, ensuring a natural and engaging conversation. You can choose the agent’s language, select or clone a voice, and fine-tune aspects like response timing, expressiveness, and stability.

Key features include:

  • Selecting a language and adjusting the Voice Model for non-English options.
  • Choosing a voice from available options or importing a custom one.
  • Refining how the agent responds using settings like Patience Level, Stability, Style Exaggeration, and Similarity.
  • Optimizing latency to balance voice quality and response speed.
  • Using Voice Prompting to adjust intonation, pacing, and emotion.
  • Resetting all voice settings (except Language and Voice) to system defaults.

These settings provide a foundation for configuring your agent’s language, voice, and tonality.

Voice configuration

1. Click the Voice tab.

2. From the Voice tab, you can customize:

a. Language: Select your agent’s language. Click Advanced Settings to change the Voice Model for languages other than English.

b. Voice: Click Edit to choose your agent’s voice and accent, import from a library, or clone a voice.

c. Patience Level: Set how long your agent waits before responding.

d. Stability: Adjust between a more expressive or steady voice.

e. Style Exaggeration: Enhance the original voice’s expressiveness.

f. Similarity: Control how closely the AI matches the original voice.

g. Optimize Latency: Balance between voice quality and response speed.

h. Voice Prompting: Add descriptors to shape intonation, pacing, and emotion.

i. Reset to Defaults: Restore all settings (except Language and Voice) to their original system defaults.

Scroll down to see more options.

Automatic fallback

Our primary voice provider is Elevenlabs. If Elevenlabs experiences any issues, your call will seamlessly switch to an alternative voice provider to prevent interruptions. In rare cases, you may notice a slight change in the agent’s voice, but the conversation will continue without dropping or interruption.