Edit text in current window by giving voice instructions to GPT-3.
- sign up for openai API
- email and phone number verification is required
- add
OPENAI_API_KEY
to.env
file - unpack
vosk
english language model (e.g. vosk-model-en-us-0.22) to the application directory- if 1.8GB model is used then the app takes ~5.5GB in RAM
pipenv install
pipenv shell
sudo -E python main.py --model vosk-model-en-us-0.22 --device 5
- running as root is required for keyboard hooks to work
-E
flag is required to passOPENAI_API_KEY
to the root shell- specify your own model derictory
- you may need to experiment with sound input device ids (starting from 0)
- some useful commands to debug devices:
pactl list sources | grep 'Name: '
,arecord -l
- some useful commands to debug devices:
- put the cursor into the text input that supports
ctrl+a
,ctrl+c
andctrl+v
- press
alt gr
(right alt key) to copy text and dictate edit instructions- hold
esc
to cancel if needed
- hold
- the edited result will be pasted back into the input