The preprocessing phase involved creating spectrograms from audio files using Librosa and preprocessing images with OpenCV. A TensorFlow CNN model was utilized, with hyperparameter optimization applied to achieve over 90% accuracy in classifying urban sounds. The dataset used for this project can be accessed here.
Successfully completed the Urban Sound Classification project using CNN, prepared by teammates in Group 32. This project was developed for the Deep Learning Bootcamp organized by Global AI Hub, sponsored by Koç Holding and Aygaz.