Text-to-Speech on Arabic Speech 59 speakers (test)
1.42WERFanar 2.0 TTS model
Evaluation results
| Method | Links | |
|---|---|---|
| Fanar 2.0 TTS modelBackbone=F5-TTS, Training steps=600k, Architecture=Diffusion Transformer (DiT)2026.03 | 1.42 | |
| ElevenLabs TTSAccess=API2026.03 | 1.51 | |
| XTTSTuning=Finetuned on Arabic data2026.03 | 1.55 |