DSAP course project, Barcelona School of Telecommunication Engineering (ETSETB), UPC Authors: Anatolii Skovitin, Francesco Maccantelli Year: 2023/2024
This repository contains all the code created for this project, which is mainly divided into two sections.
Are the files used in order to create the MEL-spectrograms from the audio files: process_dataset_img.py This script is used to create images for the CNN approach. preocess_dataset_data.py This script creates the data used for the hybrid approach.
main_alexnet_googlenet.py This script is used to train the CNN approach. It allows the user to choose between two models for training: AlexNet and GoogLeNet. main_cnn_rnn.ipynb The CNN-RNN approach is trained using this script.