eduardem/xtts-v2-romanian-tts

XTTS-v2 Romanian TTS

Romanian text-to-speech using a fine-tuned XTTS-v2 model with 5 voices.

• —Model: eduardem/xtts-v2-romanian
• —Quality: 6.3% WER (Whisper large-v3)
• —Training: ~150 hours of Romanian speech, 50 epochs
• —Voices: Costel, Georgel, Dorel, Mărioara, Lăcrămioara