wavu/buzzasr-korean-ggml

BuzzASR Korean — ggml

ggml conversions of BuzzASR/korean, a Korean fine-tune of openai/whisper-large-v3, for use with CrispASR and other whisper.cpp-compatible runtimes.

All credit for the model goes to the BuzzASR authors — see the original model card for training details and benchmarks (FLEURS + Common Voice combined CER 4.61%, vs 5.72% for Whisper-large-v3 zero-shot).

Files

File	Size	Notes
ggml-buzzasr-korean-f16.bin	3.10 GB	Full precision (f16), straight conversion
ggml-buzzasr-korean-q8_0.bin	1.66 GB	8-bit, near-lossless
ggml-buzzasr-korean-q5_0.bin	1.08 GB	5-bit, smallest

Usage

Download a file:

hf download wavu/buzzasr-korean-ggml ggml-buzzasr-korean-q8_0.bin --local-dir .

Transcribe with CrispASR:

crispasr -m ggml-buzzasr-korean-q8_0.bin -l ko -f audio.wav

Always pass -l ko. The model is fine-tuned for Korean only.

License

MIT, inherited from BuzzASR/korean.

Citation

If you use this model, please cite the BuzzASR paper: arXiv:2609.09554 (Findings of EMNLP 2026).