Urban Sounds Classification

Koç Holding Deep Learning Bootcamp by GlobalAIHub

This project uses “UrbanSounds8K” dataset to train the CNN model. You can turn .wav files into spectograms using Librosa library.

Preprocessing

Unprocessed Image                          Processed Image

Results

Test accuracy: 0.87693190574646

Loss graph                                 Accuracy graph