PlotweaverAI/f5-tts-yoruba-vocab-modified

F5-TTS Yoruba

Fine-tuned F5-TTS (F5TTS_v1_Base) for Yoruba text-to-speech with zero-shot voice cloning.

Files

• model_last.pt — EMA-pruned checkpoint for inference
• vocab.txt — custom vocabulary (base + Yoruba graphemes / combining tone marks)

Inference

from f5_tts.api import F5TTS
m = F5TTS(model="F5TTS_v1_Base", ckpt_file="model_last.pt", vocab_file="vocab.txt", use_ema=True)
wav, sr, _ = m.infer(ref_file="ref.wav", ref_text="<ref transcript>", gen_text="<yoruba text>")

If outputs are noise, retry with use_ema=False.

Notes

• Text must be NFC-normalised, matching the training frontend.
• License: verify against the base model before commercial use.