Invertible DNN-based nonlinear time-frequency transform for speech enhancement

TL;DR

This paper introduces an invertible DNN-based nonlinear time-frequency transform for speech enhancement, enabling end-to-end training while preserving perfect reconstruction, thus combining interpretability with deep learning capabilities.

Contribution

It proposes a novel invertible nonlinear T-F transform based on DNNs that ensures perfect reconstruction, improving interpretability and effectiveness in speech enhancement.

Findings

The proposed transform achieves perfect reconstruction.

Enhanced speech quality demonstrated in experiments.

Outperforms traditional linear T-F transforms.

Abstract

We propose an end-to-end speech enhancement method with trainable time-frequency~(T-F) transform based on invertible deep neural network~(DNN). The resent development of speech enhancement is brought by using DNN. The ordinary DNN-based speech enhancement employs T-F transform, typically the short-time Fourier transform~(STFT), and estimates a T-F mask using DNN. On the other hand, some methods have considered end-to-end networks which directly estimate the enhanced signals without T-F transform. While end-to-end methods have shown promising results, they are black boxes and hard to understand. Therefore, some end-to-end methods used a DNN to learn the linear T-F transform which is much easier to understand. However, the learned transform may not have a property important for ordinary signal processing. In this paper, as the important property of the T-F transform, perfect reconstruction is considered. An invertible nonlinear T-F transform is constructed by DNNs and learned from data so that the obtained transform is perfectly reconstructing filterbank.