Speech enhancement and speaker verification with convolutional neural networks

Abstract

Convolutional neural networks have been used with great success for image processing tasks. The convolution filters can act as powerful feature detectors in images to find edges and shapes. Time domain speech signals are one dimensional, but frequency domain information can be viewed in the same way as an image, allowing one to use two-dimensional convolutional networks for tasks such as speech enhancement. In this paper, recently reported deep neural network preprocessing methods used for dereverberation are examined to explore their ability to remove noise and or reverberation. A convolutional network is then introduced and compared to the previous methods to explore the relative strengths and weaknesses of each approach. Comparisons are made through experimentation in the context of speech quality and speaker identification tasks.