Building an open source talking plush with a voice you can talk to.
Our kids should grow up in a household full of hackable droids they can play with, work on, and learn from.
-
Copy the
.env.example
to.env
and fill in real values. -
Download server-side models and put them inside of
openduck-py/models
(TODO: share models.)
-
Install espeak (
brew install espeak
on Mac OS orsudo apt-get install espeak-ng
on Debian Linux).- There may be other environment variables you need to set on Mac OS. I had to:
export PHONEMIZER_ESPEAK_LIBRARY=/opt/homebrew/Cellar/espeak/1.48.04_1/lib/libespeak.dylib
.
- There may be other environment variables you need to set on Mac OS. I had to:
-
pip install -r openduck-py/requirements.txt
cd openduck-py
uvicorn openduck_py.routers.main:app --reload --env-file ../.env
docker-compose up
cd clients/simple
# Lighter-weight requirements
pip install -r requirements.txt
python simple_bot.py --record
This will start up streamlit on port 8501 so make sure that port is forwarded if you are runnin on ssh.
Install streamlit
pip install streamlit
Run streamlit
cd openduck-py
streamlit run observability.py
Alternative is to use Dockerfile
- The Simple Python client sometimes has strange audio playback bugs. You can try restarting your OS's audio services, e.g.
sudo pkill coreaudiod
on Mac OS.