Copy page

Voice (TTS)

Cartesia

Cartesia's Sonic-3 model delivers the lowest TTS latency at 37ms average, with 16 supported languages and expressive emotion control.

Requires: Cartesia API key in Settings → Credentials

Models

Model	Description
sonic-3	Recommended — fastest, multilingual

Configuration

Code
{
  "voice": {
    "provider": "cartesia",
    "voice_id": "a0e99841-438c-4a64-b679-ae501e7d6091",
    "model": "sonic-3",
    "speed": "normal",
    "emotion": ["Calm"],
    "volume": null,
    "language": "en"
  }
}

Fields

Field	Type	Default	Description
voice_id	string	required	Cartesia voice ID
model	string	sonic-3	TTS model
speed	preset or float	null	slowest/slow/normal/fast/fastest, or -1.0 to 1.0
emotion	array of strings	null	Up to 3 emotion tags blended
volume	float (0.5–2.0)	null	Volume multiplier
language	string	"en"	Language code (required)

Languages

Cartesia supports 16 languages: en, es, fr, de, it, pt, nl, zh, ja, hi, ar, ko, pl, tr, uk, vi

Emotions

Apply emotion tags to adjust speaking style:

Code
{ "emotion": ["Happy", "Calm"] }

Available emotions: Happy, Excited, Enthusiastic, Elated, Calm, Curious, Content, Peaceful, Sympathetic, Sad, Angry, Frustrated, Disgusted, Confused, Anxious, Nostalgic, Determined, Confident, and more (52 total).

Finding Voices

Code
curl "https://api.neuratel.ai/v1/voices?provider=cartesia&language=en" \
  -H "Authorization: Bearer nk_live_your_api_key"

Related

• ElevenLabs — most expressive, 71ms
• Phantom Speech — no API key, Arabic-native