Instructions to use EspenTh/munin-whisper-coreml with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- WhisperKit
How to use EspenTh/munin-whisper-coreml with WhisperKit:
# Install CLI with Homebrew on macOS device brew install whisperkit-cli # View all available inference options whisperkit-cli transcribe --help # Download and run inference using whisper base model whisperkit-cli transcribe --audio-path /path/to/audio.mp3 # Or use your preferred model variant whisperkit-cli transcribe --model "large-v3" --model-prefix "distil" --audio-path /path/to/audio.mp3 --verbose
- Notebooks
- Google Colab
- Kaggle
NB-Whisper for WhisperKit (Core ML)
Core ML conversions of the National Library of Norway's NB-Whisper models, for on-device speech recognition with WhisperKit.
| Folder | Source |
|---|---|
NbAiLab_nb-whisper-base |
NbAiLab/nb-whisper-base |
NbAiLab_nb-whisper-small |
NbAiLab/nb-whisper-small |
Converted with whisperkittools (whisperkit-generate-model). The weights are unchanged apart from the Core ML conversion. Each folder also contains NB-Whisper's tokenizer.json, so WhisperKit does not need to download a tokenizer separately.
let folder = try await WhisperKit.download(variant: "nb-whisper-base", from: "<this repo>")
let kit = try await WhisperKit(WhisperKitConfig(modelFolder: folder.path))
License and credit
The models are made by the National Library of Norway (Nasjonalbiblioteket, AI-Lab) and licensed under Apache 2.0. This conversion is licensed the same way. Cite NB-Whisper as described on the original model pages.
- Downloads last month
- 16