Speech enhancement based on Bayesian decision and spectral amplitude estimation

Abstract

In this paper, a single-channel speech enhancement method based on Bayesian decision and spectral amplitude estimation is proposed, in which the speech detection module and spectral amplitude estimation module are included, and the two modules are strongly coupled. First, under the decisions of speech presence and speech absence, the optimal speech amplitude estimators are obtained by minimizing a combined Bayesian risk function, respectively. Second, using the obtained spectral amplitude estimators, the optimal speech detector is achieved by further minimizing the combined Bayesian risk function. Finally, according to the detection results of speech detector, the optimal decision rule is made and the optimal spectral amplitude estimator is chosen for enhancing noisy speech. Furthermore, by considering both detection and estimation errors, we propose a combined cost function which incorporates two general weighted distortion measures for the speech presence and speech absence of the spectral amplitudes, respectively. The cost parameters in the cost function are employed to balance the speech distortion and residual noise caused by missed detection and false alarm, respectively. In addition, we propose two adaptive calculation methods for the perceptual weighted order p and the spectral amplitude order β concerned in the proposed cost function, respectively. The objective and subjective test results indicate that the proposed method can achieve a more significant segmental signal-noise ratio (SNR) improvement, a lower log-spectral distortion, and a better speech quality than the reference methods.

Speech Enhancement with Variance Constrained Autoencoders

“…The MUSHRA test consisted of 20 units, each unit presented the labelled clean audio file as a reference, and six other audio files with hidden labels. The six hidden audio files are: 1) The original noisy signal, 2) The enhanced noisy signal produced by SE-GAN, SE-WaveNet and SE-VCAE, 3) The clean audio file (this is the hidden reference), 4) The noisy speech signal but with an SNR 5dB less than the signal being enhanced [39] (this is the anchor). For each of the 20 units, the unlabelled audio files were in a randomised order.…”

Online/offline score informed music signal decomposition: application to minus one

“…The MUSHRA listening test [77] is a commonly used method for the subjective evaluation of audio quality. It does not require a huge number of participants to obtain a statistically significant result [79,80] reference. For this reason, we have used the MUSHRA listening test to evaluate the subjective quality of the SS results of the URMP database.…”

A Power Efficient Scheme for Speech Controlled IoT Applications

“…Speaker independence in ASR should only be applied if the nature of application demands it. [18] B. Proposed Solution The proposed scheme is an IoT implementation at constrained embedded level where sensor nodes are provided local IP address while major routing nodes are provided full-fledged internet protocol support which also solves the security issue.…”