This folder contains codes and a project report of my BSc Computing final year project. In this project, A Unet model was created to perform music source separation and separates a music into vocals and accompaniments. The implementation is based on SPECTROGRAM-CHANNELS U-NET: A SOURCE SEPARATION MODEL VIEWING EACH CHANNEL AS THE SPECTROGRAM OF EACH SOURCE

The scripts and a sample notebook have been uploaded to this repository. To access the full project folder which contains the train and test datasets, trained models and scripts, use the unzipped folder here. Make sure you download the entire Final_Year_Project folder and pass the path as the argument of Config_Handler.init() ( In order to run the scripts and train a model, you need to call the init method of Config_Handler class by passing the appropriate path that points to the unzipped folder. Have a look at the notebook in the repo for an example)