Speech Enhancement In Multiple-Noise Conditions using Deep Neural Networks

TL;DR

This paper addresses speech enhancement in complex real-world environments with multiple simultaneous noises by proposing DNN-based strategies, including psychoacoustic model-based training, to improve speech quality.

Contribution

It introduces novel DNN training strategies tailored for multi-noise conditions and incorporates psychoacoustic models for enhanced speech restoration.

Findings

DNN strategies significantly improve speech clarity in multi-noise environments

Psychoacoustic model-based training enhances noise suppression effectiveness

Approaches outperform traditional single-noise enhancement methods

Abstract

In this paper we consider the problem of speech enhancement in real-world like conditions where multiple noises can simultaneously corrupt speech. Most of the current literature on speech enhancement focus primarily on presence of single noise in corrupted speech which is far from real-world environments. Specifically, we deal with improving speech quality in office environment where multiple stationary as well as non-stationary noises can be simultaneously present in speech. We propose several strategies based on Deep Neural Networks (DNN) for speech enhancement in these scenarios. We also investigate a DNN training strategy based on psychoacoustic models from speech coding for enhancement of noisy speech