Voices

Voice213 has 30 voices, from calm and gentle to deep and bold. Every voice speaks Algerian, Tunisian and Libyan darija. Press play to hear a voice, then copy its ID.

All 30 voices work on every plan, with every model and in both formats of POST /tts. GET /tts/voices lists them.

​Use a voice

Pass the voice's ID as voice in a text to speech request, with the model you want:

{
  "text": "مرحبا بيكم، اليوم نحكيولكم على منتوج جديد.",
  "model": "voice213-pro",
  "voice": "mahmoud"
}

or as voiceId in the inline format, to get the audio inside the response:

{
  "text": "مرحبا بيكم، اليوم نحكيولكم على منتوج جديد.",
  "model": "voice213-lite",
  "voiceId": "amine"
}

• Voice IDs are lowercase, but any capitalization works. An unknown voice is refused with 400 UNKNOWN_VOICE as voice, and with 400 Invalid voice ID as voiceId.
• When a request names no voice, the API uses Nariman.
• Every voice works with every model and every dialect.

​All voice IDs

​Choosing a voice

• Try it with your real script. A voice that shines in a promo may be too energetic for a tutorial.
• Match your audience and product. A calm, warm voice suits care and family products; a deep, confident one suits tech, cars and premium goods; a bright, cheerful one suits fashion and deals.
• Pair contrasting voices when a video has two speakers: a female and a male voice, or a soft and a deep one.
• Stay consistent. Use the same voice across a campaign or a store so customers recognize it.