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.
Highlights
Speech enhancement could improve the quality of noisy speech, which results in a broad range of applications, such as mobile speech communication, robust speech recognition, aids for the hearing impaired, and so on
In order to solve the aforementioned problems, we propose a single-channel speech enhancement method based on Bayesian decision and spectral amplitude estimation (BDSAE), in which the importance of the speech detection and estimation for speech enhancement are jointly considered
By taking into account both detection and estimation errors, we propose a combined cost function, in which the cost parameters are used to balance the speech distortion and residual noise caused by missed detection and false alarm, respectively
Summary
Speech enhancement could improve the quality of noisy speech, which results in a broad range of applications, such as mobile speech communication, robust speech recognition, aids for the hearing impaired, and so on. Speech signal is present only in some frames based on short-time analysis, and only some frequency bins contain significant energy in each frame. This means that the spectral amplitude of speech signal is generally sparse. The existing speech enhancement methods do not take the sparse characteristics into consideration and often only focus on estimating the spectral amplitude rather than detecting the speech presence or speech absence. Under the assumption of speech presence uncertainty, Ephraim and Malah derived a shorttime spectral amplitude (STSA) estimator [5] by applying speech presence uncertainty to the MMSE method, which
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call