whisper
约 1 分钟阅读
标签:ASR
- openai/whisperGitHubopenai/whisperopenai/whisper · HN · https://openai.com/blog/whisper/ · 参考 · https://github.com/Starmel/OpenSuperWhisper · 模型越大越准确 · 根据 CPU 不同,可能一般只能跑 base 或 small笔记:whisper
- HNHacker NewsHacker News #32927360Hacker News
- openai.com/blog/whisperLinkopenai.com/blog/whisperopenai/whisper · HN · https://openai.com/blog/whisper/ · 参考 · https://github.com/Starmel/OpenSuperWhisper · 模型越大越准确 · 根据 CPU 不同,可能一般只能跑 base 或 small笔记:whisper
- 参考
git clone https://github.com/ggerganov/whisper.cpp
cd whisper.cpp
make -j
./models/download-ggml-model.sh base.en
./main -f samples/jfk.wav -m models/ggml-base.en.bin -pc
# 国内能直接下载 ~150K/s
# 代理能上 5MB/s
# tiny.en,tiny,base.en,base,small.en,small,medium.en,medium,large-v1,large
curl -LOC- https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-base.en.bin
curl -LOC- https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-large.bin
# samples
curl -Lo samples/hp0.ogg https://upload.wikimedia.org/wikipedia/en/d/d4/En.henryfphillips.ogg
ffmpeg -i samples/hp0.ogg -ac 1 -ar 16000 samples/hp0.wav
./main -f samples/hp0.wav -pc -t $(nproc)- 模型越大越准确
- 根据 CPU 不同,可能一般只能跑 base 或 small
- hf 下载域名 cdn-lfs.huggingface.co
- ggerganov/whisper.cpp/modelsGitHub Fileggerganov/whisper.cpp/models
owhisper
brew install fastrepl/hyprnote/owhisper
owhisper pull whisper-cpp-base-q8 # 78MB
owhisper run whisper-cpp-base-q8- HF:ggerganov/whisper.cppHuggingFaceHF:ggerganov/whisper.cpphuggingface.co · huggingface.co/ggerganov/whisper.cpp
- HF:onnx-community/moonshine-base-ONNXHuggingFaceHF:onnx-community/moonshine-base-ONNXhuggingface.co · huggingface.co/onnx-community/moonshine-base-ONNX
- Models fastrepl/hyprnote/owhisper/owhisper-model/src/lib.rsGitHub Filefastrepl/hyprnote/owhisper/owhisper-model/src/lib.rs
关联信息
反向链接和本文引用的外部资料。
反向链接
- AI Awesome笔记 · whisper
Dev Awesome · LLM Awesome · ML Awesome · Stable-Diffusion Awesome · Voice · jianchang512/clone-voice · FunAudioLLM/CosyVoice
- Voice笔记 · Whisper
FunASR · Whisper · Kaldi · 声学模型+发音词典+语言模型 · | abbr. | stand for | mean | · | STT | Speech To Text | 语音转文本 | · | TTS | Text To Speech | 文本转语音 |
References
GitHub
5 条- fastrepl/hyprnote/owhispergithub.com/fastrepl/hyprnote/tree/main/owhisper
- fastrepl/hyprnote/owhisper/owhisper-model/src/lib.rsgithub.com/fastrepl/hyprnote/blob/main/owhisper/owhisper-model/src/lib.rs
- ggerganov/whisper.cpp/modelsgithub.com/ggerganov/whisper.cpp/tree/master/models
- openai/whispergithub.com/openai/whisper
- Starmel/OpenSuperWhispergithub.com/Starmel/OpenSuperWhisper
Hacker News
1 条- Hacker News #32927360HN #32927360
HuggingFace
2 条- HF:ggerganov/whisper.cpphuggingface.co/ggerganov/whisper.cpp
- HF:onnx-community/moonshine-base-ONNXhuggingface.co/onnx-community/moonshine-base-ONNX
其他外链
1 条- openai.com/blog/whisperopenai.com/blog/whisper