Linh220905/whisper-medium-vietnamese

Whisper Medium - Vietnamese ASR (Epoch 1)

Fine-tuned openai/whisper-medium on Vietnamese speech dataset.

Training Details

• Base model: openai/whisper-medium
• Epochs: 1
• Validation Metric:
  • val_hard WER: ~13.36% (Score B: 86.64)
  • val_clean WER: ~1.66% (Score B: 98.34)

Usage

from transformers import pipeline

pipe = pipeline("automatic-speech-recognition", model="Linh220905/whisper-medium-vietnamese")
result = pipe("sample.wav")
print(result["text"])