Supervised and Unsupervised Speech Enhancement Using Nonnegative Matrix Factorization

Abstract

Abstract-Reducing the interference noise in a monaural noisy speech signal has been a challenging task for many years. Compared to traditional unsupervised speech enhancement methods, e.g., Wiener filtering, supervised approaches, such as algorithms based on hidden Markov models (HMM), lead to higher-quality enhanced speech signals. However, the main practical difficulty of these approaches is that for each noise type a model is required to be trained a priori. In this paper, we investigate a new class of supervised speech denoising algorithms using nonnegative matrix factorization (NMF). We propose a novel speech enhancement method that is based on a Bayesian formulation of NMF (BNMF). To circumvent the mismatch problem between the training and testing stages, we propose two solutions. First, we use an HMM in combination with BNMF (BNMF-HMM) to derive a minimum mean square error (MMSE) estimator for the speech signal with no information about the underlying noise type. Second, we suggest a scheme to learn the required noise BNMF model online, which is then used to develop an unsupervised speech enhancement system. Extensive experiments are carried out to investigate the performance of the proposed methods under different conditions. Moreover, we compare the performance of the developed algorithms with state-of-the-art speech enhancement schemes using various objective measures. Our simulations show that the proposed BNMF-based methods outperform the competing algorithms substantially.

A Study on the Relationship between the Intelligibility and Quality of Algorithmically-Modified Speech for Normal Hearing Listeners

“…Unlike traditional speech enhancement techniques (e.g., [1][2][3][4][5]), which focus on dealing with the noise-corrupted speech signal (i.e., speech-plus-noise mixture) and on removing background noise from the signal to achieve better listening experiences for listeners, these speech modification algorithms aim to alter the original clean speech signal so that the intelligibility may be preserved even when listened to in non-ideal listening conditions, in which background masking sources may exist. While the majority of modification algorithms operate in the frequency domain, such as enhancing frequency components which are important to speech intelligibility in noise [6][7][8] and boosting certain spectral regions based on optimising objective intelligibility metrics [9][10][11][12], other algorithms make changes in the time domain, including introducing pauses into speech and speeding up or slowing down part of the speech to avoid a temporal clash between the speech and masker [10,13].…”

An EM algorithm for joint source separation and diarisation of multichannel convolutive speech mixtures

“…(7) corresponds to a non-negative matrix factorization (NMF) model placed on the F × L matrix of variances of the source coefficients; a now common practice in audio signal processing, e.g. [16,17,18].…”

Improved Sparse NMF based Speech Enhancement Method with Deep Neural Network

“…Statistical properties based approaches like minimum mean squared error (MMSE) estimation and optimally-modified log-spectral amplitude (OM-LSA) which could take human hearing properties into account and reduce speech distortion and residual noise in some extent [1] . Recent years, supervised learning methods have achieved significant development in speech signal processing [2][3] . As a famous method which could mine implicit local representation in non negative data, non-negative matrix factorization (NMF) uses no-negative linear combination to separate clean and noise signal from noisy speech.…”