NOTICE: Python Ver 3.10 & MACOS ONLYONLY

This project streams ASL hand detection output directly to OBS using NDI and MediaPipe Hands.

git clone https://github.com/gamerwaves/HandSpeak.git

cd HandSpeak

python3 -m venv venv

source venv/bin/activate  # On Windows: venv\Scripts\activate

Note:
If you're not using bash, the activation command may vary..

pip install -r requirements.txt

• Download the latest version of OBS Studio from https://obsproject.com/download

• Download from: https://obsproject.com/forum/resources/distroav-network-audio-video-in-obs-studio-using-ndi®-technology.528/
• DistroAV is a lightweight NDI display/preview and routing tool.
• Use the installation instructions on the download page for your os.

Note:
On Linux, you also need to install ndi-sdk using your distribution's package manager.

python main.py

This will start a TKinter window that will start MediaPipe hand tracking and send frames over NDI.

• In OBS, double-click NDI Source under Sources.
• Click the small dropdown arrow beside the source name.
• Select the NDI stream titled ASL-Detection-NDI.

• You can now use this virtual camera in Zoom, Google Meet, Discord, etc. Use as a regular camera input..

• Make sure OBS and the Python script are on the same network (localhost is fine).
• DistroAV requires the latest OBS version for it to work properly.

• Built with MediaPipe, OpenCV, and NDI.
• OBS integration via obs-websocket and obs-ndi.
• README.md by ChatGPT :)