Ghazouaniwala/silma-tts-derja

SILMA TTS — Tunisian Derja fine-tune

Fine-tune of silma-ai/silma-tts (F5-TTS v1.1.7 architecture) on the LinTO Tunisian audio dataset to produce a Tunisian Derja voice with Arabic/French code-switching.

Usage

Load with the F5-TTS v1.1.7 / SILMA pipeline, using model.pt as the checkpoint, vocab.txt as the tokenizer vocab, and config.yaml for the model config.

Training data

LinTO DataSet Audio for Arabic Tunisian (subsets: ['OneStory', 'ApprendreLeTunisien', 'TunSwitchTO', 'MASC']), resampled to 24 kHz, clips 1.0-15.0s.

Attribution (required — CC BY 4.0)

This model was trained on the LinTO Tunisian dataset. Please cite:

@misc{linagora2024Linto-tn,
  title  = {LinTO Audio and Textual Datasets to Train and Evaluate Automatic Speech Recognition in Tunisian Arabic Dialect},
  author = {Hedi Naouara and J'er\^ome Louradour and Jean-Pierre Lorr'e},
  year   = {2025}, eprint = {2504.02604}, archivePrefix = {arXiv}, primaryClass = {cs.CL}
}

Base model: SILMA TTS (Apache-2.0). Dataset: LinTO (CC BY 4.0).

Responsible use

Voice cloning requires documented speaker consent. Do not use for deception, fraud, misinformation, or impersonation. Disclose AI-generated audio.