Your agent's voice is the first thing users hear. Vozzo provides a wide library of natural-sounding voices across multiple languages and accents, with support for custom voice cloning.
How Voice Selection Works
Voice availability depends on two chained selections:
- Model (
vz_model_id) — determines which languages are available - Language (
agent_language) — determines which voices and accents are available - Voice (
voice_id) — the specific voice persona
This dependency chain ensures you always select a voice that is compatible with your chosen AI pipeline.
Step 1 — Fetch Master Data
cURL
Code
Note:
vz_model_idin the language object is an array — a language may be supported by multiple models.
Step 2 — Fetch Voices by Language
cURL
Code
Voice Object Reference
| Field | Type | Description |
|---|---|---|
voice_id | integer | Use this ID in agent create/update |
name | string | Display name of the voice |
gender | string | Male, Female, Neutral |
country | string | Country of origin / accent |
provider | string | Underlying TTS provider |
is_cloned | boolean | Whether this is a cloned voice |
sample_recording | string | URL to audio preview |
Setting Voice on an Agent
cURL
Key fields
| Field | Type | Description |
|---|---|---|
agent_language | string | Must match a language from master data |
agent_accent | string | Accent identifier (e.g. "American", "British") |
voice_id | integer | ID from the voices endpoint |
Supported Languages (Sample)
Language | Orbit | Quantum |
|---|---|---|
| English | ✓ | ✓ |
| Hindi | — | ✓ |
| Spanish | — | ✓ |
| French | — | ✓ |
Fetch the live list via /master/master-data — this table reflects the current production state.
Voice Cloning
Vozzo supports cloning your own voice using ElevenLabs or Cartesia as the underlying provider.
Code
Cloned voices appear in the voices list with "is_cloned": true and can be assigned to any agent.
Best Practices
🎯 Match language to your audience — using a native-language voice dramatically improves user trust.
🔊 Preview before deploying — always listen to the
sample_recording URL before assigning a voice.
⚡ Orbit supports fewer languages — if you need a specific language, check availability before choosing Orbit.
🎙️ Voice cloning — provide at least 60 seconds of clean audio for best clone quality.
Last updated: March 2026 | API Version: v1

