Deep Learning for Audio

Implementation of State-of-the-arts in Audio Event Classification, Sound Event Detection (SED) and even more...in Pytorch

1. Audio Classification

CNN-1D Solution:

Build a conv1d model using directly raw waves from audio files

CNN-2D Solution:

Using log-mel spectrogram as features
CNN-2D as core model, based on image classification task

Benchmark Dataset

The benchmark audio dataset was obtained from Kaggle

2. Sound Event Detection

TODO

References