First, clone this repo and install whisper.cpp in this directory
git clone https://github.com/aspctu/substrate.git
cd substrate
git clone https://github.com/ggerganov/whisper.cpp
cd whisper.cpp
make
Second, install other reqs
brew install portaudio ffmpeg
poetry install
poetry shell
python3 main.py
Transcriptions will land in output.txt in real-time unless invoked with output flag
python3 main.py --output="path/to/my/file"
By default, this uses the base.en model. You can use smaller / larger models as you wish.
First, download the model you want
cd whisper.cpp
make [size you want, e.g large]
Second, update the variable WHISPER_MODEL in constants.py
- WHISPER_MODEL = "base.en"
+ WHISPER_MODEL = "large"
❤️ to whisper.cpp and speech_recognition