ggml-PhoWhisper-tiny

VinAI PhoWhisper-tiny converted to ggml for whisper.cpp / browser WASM.

Files

File	Size	Notes
ggml-phowhisper-tiny-q8_0.bin	~42 MiB	Recommended — Q8_0 quantized, near-fp16 quality
ggml-phowhisper-tiny.bin	~74 MiB	Full fp16 (reference / max quality)

ACut web captions use the Q8_0 file by default.

Cite

@inproceedings{PhoWhisper,
  title     = {{PhoWhisper: Automatic Speech Recognition for Vietnamese}},
  author    = {Thanh-Thien Le and Linh The Nguyen and Dat Quoc Nguyen},
  booktitle = {Proceedings of the ICLR 2024 Tiny Papers track},
  year      = {2024}
}