Audio Classification using LSTM

Classification of Urban Sound Audio Dataset using LSTM-based model.

Requirements

- pytorch==1.0.1
- scipy==1.2.0
- torchvision==0.2.1
- pandas==0.24.1
- numpy==1.14.3
- torchaudio==0.2
- librosa==0.6.3
- pydub==0.23.1

Steps to follow for testing on your Test Data

• Create a folder named data/test in the current directory which will contain all the '.wav' files that are to be tested.

• Download 'bestModel.pt' from this Link and place in the current directory.

• Run the following commands:

python preprocess.py
python eval.py

• A csv file named 'test_predictions.csv' will be generated in the current directory containing all the test files along with their corresponding predicted labels.

Citation

In case you find any of this useful, consider citing:

@misc{audio-classification-using-LSTM,
  author = {Shagun Uppal, Anish Madan, Sarthak Bhagat},
  title = {sarthak268/Audio_Classification_using_LSTM},
  url = {https://github.com/sarthak268/Audio_Classification_using_LSTM},
  year = {2019}
}

Team

• Anish Madan
• Sarthak Bhagat
• Shagun Uppal