PyTorch conversion of Google's VGGish model used for extracting audio features for AudioSet. (https://github.com/tensorflow/models/tree/master/research/audioset) - View it on GitHub