Voice (TTS)
ElevenLabs
ElevenLabs provides industry-leading voice quality with extensive customization. Average TTS latency: 71ms.
Requires: ElevenLabs API key in Settings → Credentials
Models
| Model | Description |
|---|---|
eleven_flash_v2_5 | Recommended — fast, high quality |
Configuration
Code
Fields
| Field | Type | Default | Description |
|---|---|---|---|
voice_id | string | "gHu9GtaHOXcSqFTK06ux" | ElevenLabs voice ID |
model | string | eleven_flash_v2_5 | TTS model |
stability | float (0–1) | 0.5 | Higher = more consistent, less expressive |
similarity_boost | float (0–1) | 0.75 | Higher = closer to original voice character |
style | float (0–1) | null | Speaking style intensity |
speed | float (0.25–4.0) | 1.0 | Speech rate (recommended range: 0.8–1.2) |
use_speaker_boost | bool | null | Enhanced similarity to reference voice |
apply_text_normalization | auto/on/off | auto | Format numbers and abbreviations for TTS |
language | string | null | ISO language code or null for auto-detect |
Finding Voices
Browse and preview available voices in the dashboard at app.neuratel.ai/voices, or via API:
Code
Voice previews are available directly in the Agent Studio — click the play button next to any voice in the selector.
Related
- Cartesia — fastest TTS at 37ms
- Phantom Speech — no API key, Arabic-native
Last modified on