Sound classification on Urban Sound Dataset

Download the urban sound dataset here

• Extract the dataset on to a folder
  
• Install pytorch (gpu verison if you have gpu)
  
• Install the requirements
  

pip install -r requirements.txt

• Run the dataset.py file (change the path to the dataset)
  

python dataset.py

This will give the number of items in the dataset

• Then run modelcnn.py to know about the structure of the CNN model.
  
• Then start the training. (change the path to the dataset)
  

python train.py

• After the training, the model will be saved in a specified folder. Use the saved model for inferencing.
  Run
  

python inference.py

Reference: https://github.com/musikalkemist/pytorchforaudio