Mickey: Emotion Recognition in Music

Mickey is a ML based web app built to analyze and capture emotions in music using deep learning techniques. The model is powered by LSTM and GRU-based neural networks, and the application is deployed using FastAPI. It uses Librosa for audio processing and feature extraction, with Pandas and Numpy handling data manipulation.

Features

• Emotion Detection: Detect emotions in music tracks using deep learning models.
• LSTM & GRU: The emotion recognition model is built with Long Short-Term Memory (LSTM) and Gated Recurrent Units (GRU) for effective time series analysis.
• FastAPI: FastAPI is used for building the backend, providing high performance and easy API endpoints.
• Audio Processing: Librosa is used for feature extraction from audio tracks, including Mel spectrograms, chroma features, and more.
• User Interface: Jinja templates power the front-end, offering a simple yet efficient user interface for interacting with the application.

Requirements

To run the project, you'll need the following libraries:

• Pandas: For handling data in tabular form.
• NumPy: For numerical operations and data manipulation.
• TensorFlow: For building and training the LSTM and GRU-based neural network models.
• Librosa: For audio processing and extracting features from music tracks.
• FastAPI: For backend development and creating RESTful APIs.
• Jinja: For rendering HTML templates.
• Uvicorn: ASGI server for FastAPI.

You can install the required dependencies using the following command:

pip install pandas numpy tensorflow librosa fastapi jinja2 uvicorn

Installation

1. Clone the repository:

   git clone https://github.com/RijoSLal/mickey.git

2. Navigate into the project directory:

   cd mickey

3. Run the FastAPI app:

   uvicorn fast:app --reload

   The application will be available at http://127.0.0.1:8000.

Usage

1. Upload a music file: Navigate to the app's web interface and record an audio and submit.
2. Get emotion prediction: The model will process the audio file and provide an emotion classification based on the music's features and shows audio spectrogram.

Contributing

If you'd like to contribute to Mickey, feel free to fork the repository and submit a pull request. Here are a few ways you can help:

• Report issues or bugs.
• Add new features or improve existing ones.
• Improve documentation.

License

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

Feel free to reach out if you have any questions or suggestions!