eduardem/f5-tts-romanian-tts
2
F5-TTS Romanian TTS
Romanian text-to-speech using a fine-tuned F5-TTS model with 5 voices.
- Model: eduardem/f5-tts-romanian
- Quality: 5.1% WER (Whisper large-v3)
- Training: ~112 hours of Romanian speech, 100 epochs, 235K steps
- Voices: Costel, Georgel, Dorel, Mărioara, Lăcrămioara