environmental-sound-classification

This project is a classifier trained on environmental sounds. The model can classify the input sound into 10 groups. The dataset that I used in this task is UrbanSound8k. Urban Sound 8K is an audio dataset that contains 8732 labeled sound excerpts (<=4s) of urban sounds from 10 classes: air_conditioner, car_horn, children_playing, dog_bark, drilling, enginge_idling, gun_shot, jackhammer, siren, and street_music.
I used a hand-designed CNN with much fewer parameters than a Resnet18 model for feature extraction and some linear layers for classification.