Voice

---

The Voice Tab

The Voice tab shapes how your AI agent sounds and listens - defining its entire auditory experience. Here, you control how it understands speech, responds with lifelike audio, and maintains a natural flow during calls or voice interactions.

---

Language

This determines the language and accent your voice agent will use for both listening and speaking.

• Language: Choose the specific language and regional dialect (e.g., English (India)). This ensures the agent correctly interprets user speech and responds in a matching tone and vocabulary.

---

Speech-to-Text (Input)

This section governs how the agent hears and understands the user. The system uses a transcriber to convert spoken words into text before processing them through the AI model.

• Select Transcriber: Choose the Speech-to-Text engine that best suits your application.
  • Provider: The technology provider offering the transcription service (e.g., Deepgram).
  • Model: The specific transcriber model version (e.g., nova-3), determining how quickly and accurately speech is converted to text.

---

Text-to-Speech (Output)

This defines how the agent speaks back to the user. The selected voice synthesizer turns the AI’s text responses into natural-sounding speech.

• Select Voice: Pick the Text-to-Speech (TTS) engine and voice that matches your brand’s style.
  • Provider: The TTS service powering the audio output (e.g., ElevenLabs).
  • Model / Voice: The specific voice profile or tone you want the agent to use (e.g., Jeevan – Expressive Indian Voice).

---

Audio Delivery Fine-Tuning

Fine-tune how your agent sounds during conversations - balancing clarity, timing, and natural flow.

• Speed Rate: Controls how fast or slow your agent speaks. Adjust the slider to match your preferred pace (default is 1).
• Buffer Size: Controls how much audio is preloaded before playback. A higher buffer size results in smoother long responses, but may introduce a slight delay before speech begins.