Repository navigation
What's the recommended dictation approach? #143
Replies: 2 comments
|
For fully local dictation on a Mac there are two good routes: 1. whisper.cpp (same ggml stack as llama.cpp). This is the natural fit here. It runs Whisper models on Metal with no network: git clone https://github.com/ggml-org/whisper.cpp && cd whisper.cpp
sh ./models/download-ggml-model.sh base.en # or small / large-v3-turbo
brew install sdl2 # for live microphone input
cmake -B build -DWHISPER_SDL2=ON && cmake --build build -j --config Release
# live transcription from the microphone
./build/bin/whisper-stream -m models/ggml-base.en.bin --step 500 --length 5000To build it into an app, use the C API in 2. Apple's on-device speech framework. A common setup is to stream microphone audio to whisper.cpp, then feed the text into the model as the prompt, so everything (speech-to-text and the LLM) stays on the machine. |
|
Check https://github.com/cjpais/Handy by @cjpais, @saharmilis (and please share your feedback) |
Uh oh!
There was an error while loading. Please reload this page.
To add a dictation feature, to run 100% locally.
What's the recommended approach?
All reactions