Downloads · 30 days
19
51% of all-time downloads
luvox-ai/vietasr-ssl-char42-canonical-vi-ctc
vietasr-ssl-char42-canonical-vi-ctc is a automatic speech recognition model from luvox-ai. Use it when you need speech turned into text.
Standalone Vietnamese CTC ASR checkpoint. This repo includes the minimal runtime needed to load model.safetensors and run inference, so users do not need to clone the full experiment repo.
Downloads · 30 days
19
51% of all-time downloads
All-time downloads
37
Public
Parameters
65.7M
263 MB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors263 MB · 100%
From the Hugging Face model README
Standalone Vietnamese CTC ASR checkpoint. This repo includes the minimal runtime needed to load model.safetensors and run inference, so users do not need to clone the full experiment repo.
| Eval | Samples | TER | WER | CER |
|---|---|---|---|---|
| VI eval: BTN + Golden + LSVSC test | 1166 | 3.70% | 7.03% | 3.84% |
| PW733 406: 170826 + 180826 + BTN + Golden | 406 | 5.43% | 13.19% | 5.97% |
git clone https://huggingface.co/luvox-ai/vietasr-ssl-char42-canonical-vi-ctc
cd vietasr-ssl-char42-canonical-vi-ctc
export PATH="/Utilisateurs/tnguye28/tools/ffmpeg:/home/tools:$PATH"
export LD_LIBRARY_PATH="/Utilisateurs/tnguye28/Speech-LLMs/Auden/.venv/lib/python3.11/site-packages/nvidia/nvtx/lib:${LD_LIBRARY_PATH:-}"
export CUDA_VISIBLE_DEVICES=0
/Utilisateurs/tnguye28/.conda/envs/auden_cuda124/bin/python infer_ctc.py audio.wav --device cuda:0