AIGDSCMeet is a project that utilizes technologies such as FastAPI, Socket.IO, Gunicorn, PyTorch, CUDA, and Transformers to process and extract information from audio in meetings using GPU acceleration.
-
Clone the project from GitHub:
git clone https://github.com/TrMTu4898/services-ai-gdscmeet.git cd services-ai-gdscmeet
-
Run Docker Compose:
docker-compose up --build
-
Access the application at ws://127.0.0.1:5000/
APIAI/main.py
: Contains the main source code for the FastAPI, Gunicorn, Uvicorn for AI function.AIGDSCMeet/main.py
: Container the main source code for SocketIO, FastAPI, Uvicorn connect to the client at the gateway [ws://127.0.0.1:5000/].APIAI/Dockerfile
: Dockerfile to build the Docker image for the api alication.AIGDSCMeet/Dockerfile
: Dockerfile to build the Docker image for socketIO application.docker-compose.yml
: Docker Compose configuration file.APIAI/requirements.txt
: List of Python dependencies for APIAI.AIGDSCMeet/requirements.txt
: List of Python dependencies for AIDSCMeet.
- After installation and running, access ws://127.0.0.1:5000/ in your web browser.
- Instructions for use
- Perform actions on the interface to interact with the application.
-
//Initialize socket connection const socket = io.connect('ws://127.0.0.1:5000/', { auth: user_info }); //Connect to the server socket.on('connect', () => { console.log('Connected to the server!'); }); //Disconnect from the server socket.on('disconnect', () => { console.log('Disconnected from the server!'); }); //Send data to the server socket.emit('speech_to_text_result', interimResults) //Receive results from the server socket.on('speechToKeywords', (result) => { console.log('Speech to Keywords Result:', result); });
-
Ensure your computer has a GPU compatible with CUDA for optimal performance of the project.
-
To optimize GPU performance, make sure to install the NVIDIA Docker Toolkit and CUDA Toolkit correctly.
-
If you encounter issues, check the logs of the containers for detailed information.