Group university project where the effectiveness of mixup augmentation methods on the sound domain are analyzed.
data_preprocessing.ipynb - prepares required data files for provided datasets (ESC50 and US8k). Can mount Google drive to store data.
sound_classification.ipynb - presents augmentation methods and performs training with ResNet18 model. Can grab data from Google drive.
Training_sound_classification.py - performs training with ResNet18 model. Created from sound_classification.ipynb to change recorded metrics, train/test split and to be able to run outside Jupyter.