EnvNet_v1_v2_TensorFlow_Keras

An implementation of EnvNet_v1 and EnvNet_v2 in Python with TensorFlow Keras.

Train an example with ESC-50 dataset.

Requirements

• Numpy
• Scipy
• librosa (0.7+)
• TensorFlow (1.14+)

Description

The EnvNet_v1.py and EnvNet_v2.py contain the model definition and the train/val/test methods.

The EnvNet_v1_data_utils.py and EnvNet_v2_data_utils.py prepare the ESC-50 dataset for the model.

Data preparation follows the Envnet_v1 paper:

• Train and validate with a random selected window from each audio recoding
• Test on sliding windows and predict at audio recording level with probability voting
• Normalize data between -1 to 1
• Remove silent window when maximum amplitude is smaller than 0.2

Reference

EnvNet_v1:

@inproceedings{tokozume2017learning,
  title={Learning environmental sounds with end-to-end convolutional neural network},
  author={Tokozume, Yuji and Harada, Tatsuya},
  booktitle={2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  pages={2721--2725},
  year={2017},
  organization={IEEE}
}

EnvNet_v2:

@inproceedings{tokozume2017learning,
  title={Learning from between-class examples for deep sound recognition},
  author={Tokozume, Yuji and Ushiku, Yoshitaka and Harada, Tatsuya},
  journal={arXiv preprint arXiv:1711.10282},
  year={2017}
}

ESC-50:

@inproceedings{piczak2015esc,
  title={ESC: Dataset for environmental sound classification},
  author={Piczak, Karol J},
  booktitle={Proceedings of the 23rd ACM international conference on Multimedia},
  pages={1015--1018},
  year={2015}
}