zionia/whisper-small-isizulu-specaugment

Whisper Small - Combined South African Languages

This model is a fine-tuned version of openai/whisper-small on multiple South African language datasets.

Training Data

This model was trained on 2 combined datasets:

• zionia/isizulu-asr-train
• zionia/isizulu-asr-specaugment

Total training samples: 1060 Total test samples: 265

Training Details

• Training epochs: 15
• Learning rate: 1e-05
• Batch size: 16
• Best WER: 90.40%

Usage

from transformers import pipeline

pipe = pipeline("automatic-speech-recognition", model="zionia/whisper-small-isizulu-specaugment")
result = pipe("path/to/audio.wav")