Whisper fine-tuning event script to use multiple hf datasets
This works for only Mongolian for now but feel free to change
multiple_datasets.dataset_utils.keep_chars variable for preprocessing.
python train.py \
--train_datasets "mozilla-foundation/common_voice_11_0|mn|train+validation,google/fleurs|mn_mn|train+validation" \
--eval_datasets "mozilla-foundation/common_voice_11_0|mn|test" \
--whisper-size "small" \
--language "mn,Mongolian" \
--train-batch-size 16 \
--eval-batch-size 16 \
--max-steps 15000 \
--num-workers 8 \
--read-from-preprocessed \
--hf-username 'your-huggingface-name' \
--version 1# install multiple_datasets
!pip install git+https://github.com/bayartsogt-ya/whisper-multiple-hf-datasets.git
from multiple_datasets.hub_default_utils import convert_hf_whisper
model_name_or_path = 'openai/whisper-tiny'
whisper_checkpoint_path = './whisper-tiny-checkpoint.pt'
convert_hf_whisper(model, whisper_checkpoint_path)
# now transcribe
import whisper
model = whisper.load_model(whisper_model_path)
result = model.transcribe('loooong_audio_path.wav') # probably longer than 10 min? hour?
print(result['text'])
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent