This repository is dedicated to the sound classification project completed as part of the "COMP47650 - Deep Learning" module assignment at UCD in 2021. The project focuses on classifying sounds into 41 categories using the FSDKaggle2018 dataset.
After being dormant for several years, I've decided to make this repository public for educational purposes. Notably, this project diverges from its original inspiration, which was based on the work of @Vishwa07 in Keras. In this version, I have implemented the project using PyTorch.
Feel free to explore, learn, and contribute to this educational project!
While I never completed the training due to the substantial time requirement, I acknowledge that the project could benefit from further refinement. At the time of development, I lacked the knowledge on saving training progress. If you decide to use or explore this repository and encounter any bugs, issues, or potential improvements, I encourage you to fork the repository and submit a pull request. Your contributions are highly welcomed and appreciated!
pip install -r requirements.txt
Download the dataset from https://zenodo.org/record/2552860 and put it on "./dataset" folder
Run python train.py