Speech Separation

Design a robust Deep Convolutional Neural Network (CNN) architecture to extract vocals from music signals by exploiting Ideal Binary Masks (IBM) features. Trained with MUSDB18 dataset and evaluated…

A personal toolkit for single/multi-channel speech recognition & enhancement & separation.

DPRNN-based speech separation for single channel with two speakers

Deep neural network (DNN) for noise reduction, removal of background music, and speech separation

Two-talker Speech Separation with LSTM/BLSTM by Permutation Invariant Training method.

An RNN implementation in Matlab

Recurrent neural network training for noise reduction in robust automatic speech recognition