HKAB/whisper-finetune-vietnamese

Whisper finetuned on VinBigdata-VLSP2020-100h + KenLM

33
/ 100
Emerging

Integrates a custom BeamSearchWithLM decoder leveraging KenLM n-gram language models to improve decoding accuracy beyond standard greedy inference. Provides complete finetuning pipelines with distributed training support, alongside pre-trained checkpoints and notebooks for n-gram generation and comparative benchmarking against Wav2vec2 baselines. Reduces WER from 50% to 33% on Vietnamese speech through supervised finetuning on 100-hour in-domain data.

No commits in the last 6 months.

No License Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 7 / 25
Maturity 8 / 25
Community 18 / 25

How are scores calculated?

Stars

37

Forks

12

Language

Jupyter Notebook

License

Last pushed

Oct 06, 2023

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/nlp/HKAB/whisper-finetune-vietnamese"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.