You can go to dataset/esc50_dataset.md to get connected with the esc-50 audio dataset.
Methods review:
- Converting audio files to Spectrogram images: https://abizasobanaraj.medium.com/audio-spectrogram-generation-9d31ab695bcd
- Finetuning torch models: https://pytorch.org/tutorials/beginner/finetuning_torchvision_models_tutorial.html
- Creating custom dataset: https://pytorch.org/tutorials/beginner/data_loading_tutorial.html