Acapella splitter using Pytorch UNet with spectrograms. In progress, which means, it doesn't work yet and don't use it.
Pre-trained model will be on MUSDB18 dataset.

• Load Datasets
• Convert to Spectrograms
• Train Model
• Split Acapella
• Evaluate Model
• Improve Model to Minimize Loss
• Pre-Train Full MUSDB18 Model