Instructions to use doggy8088/Breeze-ASR-26-MLX with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use doggy8088/Breeze-ASR-26-MLX with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir Breeze-ASR-26-MLX doggy8088/Breeze-ASR-26-MLX
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
|
Download README.md from doggy8088/Breeze-ASR-26-MLX: direct link, hf CLI and curl.
- Browser
- Download file 981 Bytes
-
https://huggingface.co/doggy8088/Breeze-ASR-26-MLX/resolve/main/README.md
- Command line
-
hf download hf://doggy8088/Breeze-ASR-26-MLX/README.md
-
curl -L -o README.md https://huggingface.co/doggy8088/Breeze-ASR-26-MLX/resolve/main/README.md
981 Bytes
metadata
license: apache-2.0
library_name: mlx
pipeline_tag: automatic-speech-recognition
base_model: MediaTek-Research/Breeze-ASR-26
tags:
- mlx
- mlx-whisper
- whisper
- automatic-speech-recognition
- taiwanese-hokkien
- taigi
- zh
- nan
Breeze-ASR-26-MLX
這是 MediaTek-Research/Breeze-ASR-26 的 MLX 轉換版本。原始模型是以 Whisper Large-v2 為基礎微調的語音辨識模型。
使用方式
安裝 mlx-whisper:
pip install mlx-whisper
使用 CLI 轉錄音訊:
mlx_whisper audio.wav --model doggy8088/Breeze-ASR-26-MLX --language zh
或在 Python 中使用:
import mlx_whisper
result = mlx_whisper.transcribe(
"audio.wav",
path_or_hf_repo="doggy8088/Breeze-ASR-26-MLX",
language="zh",
)
print(result["text"])
轉換方式
此模型是從 Hugging Face Transformers checkpoint 轉換成 mlx-whisper 格式,權重使用 fp16。