DanceGen

DanceGen is an AI-powered dance generation API that creates dance animations based on music input. It leverages the EDGE (Editable Dance Generation) model to produce unique dance sequences synchronized with given songs. Please go to EDGE and make sure that you have the pre requisites and your device meets the project's GPU requirements before running the project

Watch user generated dances

I have stopped the GPU VM I rented, and now only the dance animations generated in August during the launch are available. Generated Dances

Front end

BileAI front end

Features

Generate dance animations from song names or audio files
Asynchronous processing for efficient handling of multiple requests
Integration with YouTube for audio retrieval
Storage of generated dance animations in AWS S3
Metadata storage in MongoDB for easy retrieval and management

Architecture

The DanceGen project consists of the following main components:

FastAPI backend for handling HTTP requests
Redis Queue for task management
EDGE model for dance generation
YouTube API integration for audio retrieval
AWS S3 for storage of generated FBX files
MongoDB for metadata storage

Prerequisites

Docker and Docker Compose
AWS account with S3 access
YouTube API credentials

Installation and Setup

Clone the repository:

git clone https://github.com/youngccame1/bile-ai-backend.git
cd bile-ai-backend

Create a .env file in the root directory with the following variables:

YOUTUBE_API_KEY=your_youtube_api_key
MONGODB_URI=your_mongodb_connection_string
REDIS_URL=redis://redis:6379
AWS_ACCESS_KEY_ID=your_aws_access_key
AWS_SECRET_ACCESS_KEY=your_aws_secret_key
S3_BUCKET_NAME=your_s3_bucket_name

Build and run the Docker containers:
```
docker-compose up --build
```

This will start the FastAPI server, Redis, and a worker to process the queue.

Usage

The API will be available at http://localhost:8000.

To generate a dance animation, send a POST request to the /api/v1/generate_dance endpoint:

import requests

response = requests.post(
    "http://localhost:8000/api/v1/generate_dance",
    json={"song_name": "Billie Jean"}
)

task_id = response.json()["task_id"]

You can then check the status of the task and retrieve the result using the task ID.

FBX Environment

If you need an AutoDesk FBX environment for other project you can pull the image from DockerHub which allows you to work with Python FBX without installing it manually.

docker pull youngccamel/bile:fbx2

.pkl to .fbx file converter API

Converter

API Documentation

Once the server is running, you can access the API documentation at http://localhost:8000/docs.

Docker Configuration

The project uses Docker for containerization. Here's an overview of the Docker setup:

Dockerfile

The Dockerfile in the root directory defines the main application container. It:

Uses Python 3.9 as the base image
Installs system dependencies (git, ffmpeg, libsm6, libxext6)
Clones the EDGE repository
Installs Python dependencies from requirements.txt
Copies the application code
Sets the command to run the FastAPI server

docker-compose.yml

We use Docker Compose to define and run multi-container Docker applications. Create a docker-compose.yml file in the root directory with the following content:

version: '3.8'

services:
  web:
    build: .
    ports:
      - "8000:8000"
    env_file:
      - .env
    depends_on:
      - redis
      - mongodb

  worker:
    build: .
    command: rq worker
    env_file:
      - .env
    depends_on:
      - redis
      - mongodb

  redis:
    image: "redis:alpine"

  mongodb:
    image: "mongo:4.4"
    volumes:
      - mongodb_data:/data/db

volumes:
  mongodb_data:

This Docker Compose file defines four services:

web: The main FastAPI application
worker: An RQ worker to process background tasks
redis: Redis server for task queue
mongodb: MongoDB database for storing metadata

Contributing

Contributions to DanceGen are welcome! Please refer to our Contributing Guidelines for more information.

License

This project is licensed under the MIT License. See the LICENSE file for details.

Acknowledgements

For any questions or support, please open an issue on the GitHub repository.

akbota-kurmangazhyyeva/bile-ai