Background voice detection program that listens for a wake word and activates transcription mode. After the transcription is finished, the GPT model responds to the transcribed text.
Repository is in active development.
set OPENAI_API_KEY=sk-...
git clone https://github.com/eryk-mazus/sigh.git
cd sigh
pip install -e .
# run:
python ./sigh/main.py --help
# no wake word detection:
python ./sigh/main.py --detect_wake_word=False
# running in background (soon):
# https://janakiev.com/blog/python-background/?ref=python-shell-commands
- add automatic transcription stopping criteria
- implement different modes (wake word, always transcribe, save to file)
- talk with local models, e.g. llama2, mistral, etc.
Lower priority:
- add video example
- colors, bold
- measure execution time