Whisper Fine-Tuning for Japanese-to-Chinese Translation

This project fine-tunes the Whisper base model to perform Japanese audio to Chinese text translation. It uses the Hugging Face Transformers library and a custom dataset aligned for speech translation.

Key Features

  • 🎯 Task: Speech translation (Japanese β†’ Chinese)
  • 🧠 Model: Whisper Base, fully fine-tuned
  • πŸ—ƒοΈ Dataset: ScreenTalk_JA2ZH
  • πŸ“Š Metrics: Evaluated using BLEU score

Evaluation Results

Achieved a BLEU score of ~0.72 on the held-out test set, indicating high translation quality.

Maintained by IT Banque | https://www.itbanque.com

Downloads last month
25
Safetensors
Model size
72.6M params
Tensor type
F32
Β·
Inference Providers NEW
This model isn't deployed by any Inference Provider. πŸ™‹ Ask for provider support

Dataset used to train Itbanque/whisper-ja-zh-base

Space using Itbanque/whisper-ja-zh-base 1