Habibi: Laying the Open-Source Foundation of Unified-Dialectal Arabic Speech Synthesis

This is an online demo for Habibi-TTS.

Several notes:

  • Use more qualified reference audio instead of default, ensure less than 12 seconds (use ✂ to clip)
  • Provide an exactly matched reference text, which ends with proper punctuation, e.g. a period
  • Ensure the audio is fully uploaded before generating
  • If any issues, try convert to WAV or MP3 format first

The terminology (three letters capitalized) does not reflect any official classification for Arabic.

Choose TTS Language
Choose TTS Model

Check to use a random seed for each generation. Uncheck to use the seed specified.

Example Prompts