Skip to content

The Voice213 API turns a written script into a natural voice-over, in Algerian, Tunisian or Libyan darija. The voices are made for advertising: by default they read with a warm, confident, persuasive delivery, and tone directions change it line by line.

​One endpoint

POST /tts is one endpoint for every voice-over: choose the model with model and the voice with voice, and change either one from one request to the next. All five models and all 30 voices are available:

{
  "text": "مرحبا بيكم في Voice213",
  "model": "voice213-pro-2",
  "voice": "nariman",
  "dialect": "algerian"
}

You get a link to an MP3 file, which is also saved to your History in the web app. To get the audio inside the response instead, as a WAV file, name the voice with voiceId rather than voice.

  • A voice-over takes up to 5,000 words, tone directions included, whatever your plan. For a longer script, split it into parts.
  • Voice-overs are billed per character at the model's price, from your API credit, with a minimum of 100 characters per request. Nothing is charged when a voice-over fails.
  • To know the price before you generate, send the text to Estimate Cost, free.

See the Text to Speech reference for every parameter.

​How the voice reads your script

  • It speaks like a local. In Algerian, ق becomes a hard "g" (قال sounds like "gal"), and everyday darija words sound natural. See Dialects.
  • It reads prices and numbers the local way. 2500 دج is read «خمسة وعشرين مية دينار», and 50% «خمسين بالمية».
  • It pronounces French and English words the way people in the Maghreb say them, not with a textbook accent.
  • It fixes common spelling mistakes as it reads, so a typo doesn't become a mispronunciation.
  • It respects diacritics. A word written with tashkeel is pronounced exactly as written.

​Writing a script that sounds great

  • Write the way people talk. Darija reads best when it's written as it's spoken, not translated from Modern Standard Arabic.
  • Keep sentences short. Commas and full stops become natural pauses.
  • Direct the key moments. Put a tone direction before the hook, the offer and the call to action.
  • Add tashkeel to ambiguous words such as names, or words that are spelled the same but said differently.
  • Try another voice or model if a take doesn't fit: Pro 1.5 gives the most natural delivery, and Pro 2 the most expressive.

Was this page helpful?