A Regression Approach to Speech Enhancement Based on Deep Neural Networks

TLDR

The authors propose a supervised deep neural network approach that learns a mapping from noisy to clean speech, offering an alternative to conventional MMSE-based noise reduction. They train a DNN regression model on a large, diverse dataset of speech–noise mixtures, augmenting it with global variance equalization, dropout, and noise‑aware training to reduce over‑smoothing and improve generalization to unseen noises. Experiments show the DNN method yields substantial objective and subjective gains over MMSE, effectively suppresses highly nonstationary noise, and generalizes from synthetic training data to real‑world recordings without producing musical artifacts.

Abstract

In contrast to the conventional minimum mean square error (MMSE)-based noise reduction techniques, we propose a supervised method to enhance speech by means of finding a mapping function between noisy and clean speech signals based on deep neural networks (DNNs). In order to be able to handle a wide range of additive noises in real-world situations, a large training set that encompasses many possible combinations of speech and noise types, is first designed. A DNN architecture is then employed as a nonlinear regression function to ensure a powerful modeling capability. Several techniques have also been proposed to improve the DNN-based speech enhancement system, including global variance equalization to alleviate the over-smoothing problem of the regression model, and the dropout and noise-aware training strategies to further improve the generalization capability of DNNs to unseen noise conditions. Experimental results demonstrate that the proposed framework can achieve significant improvements in both objective and subjective measures over the conventional MMSE based technique. It is also interesting to observe that the proposed DNN approach can well suppress highly nonstationary noise, which is tough to handle in general. Furthermore, the resulting DNN model, trained with artificial synthesized data, is also effective in dealing with noisy speech data recorded in real-world scenarios without the generation of the annoying musical artifact commonly observed in conventional enhancement methods.