Renhua Peng

A Constrained MMSE LP Residual Estimator for Speech Dereverberation in Noisy Environments

After revealing that both late reverberation and noise are additive interference components in the residual domain, this paper proposes to suppress these additive interference components by using a constrained minimum mean square error linear prediction (LP) residual estimator, where the optimal filter can be obtained by the generalized singular value decomposition. We propose to estimate the LP residuals for both late reverberation and noise continuously, which is based on the non-VAD related noise power spectral density estimator and the incessant late reverberant spectral variance estimator. The non-intrusive objective measure and the PESQ show that the proposed algorithm is better than traditional LP residual-based algorithms and spectral subtraction-based algorithms.

A Statistical Analysis of Two-Channel Post-Filter Estimators in Isotropic Noise Fields

This paper derives explicit expressions of the probability density functions of the two-channel post-filter estimators in isotropic noise fields to study their statistical properties. According to the analysis results, three methods are proposed to improve the performance of the noise filed coherence (NFC)-based post-filter estimator.

Robust Adaptive Beamforming Using Noise Reduction Preprocessing-Based Fully Automatic Diagonal Loading and Steering Vector Estimation

Diagonal loading provides a powerful and effective way to improve the robustness of the standard Capon beamformer. Several parameter-free robust adaptive beamformers (RAB) are considered in this paper. We reveal that the performances of them have somewhat degradation when the number of snapshots or that of sensors is large. To solve this problem, we emphatically study the well-known generalized linear combination-based method, the performance of which may degrade severely when the number of sensors increases, and propose a novel parameter-free technique, which is a combination of noise reduction preprocessing technique and truncated minimum mean square error criterion. As most of the parameter-free RAB techniques are very sensitive to the desired signal steering vector mismatch, this paper further proposes to construct a series connection between these RAB techniques and a steering vector estimation (SVE) method, where the SVE is implemented by a convex optimization technique. Simulation results show that the proposed method can achieve a promising performance in comparison with the competing methods.

A perceptually motivated LP residual estimator in noisy and reverberant environments

Abstract Both reverberation and additive noise can degrade the quality of recorded speech and thus should be suppressed simultaneously. Previous studies have shown that the generalized singular value decomposition (GSVD) has the capability of suppressing the additive noise effectively, but it is not often applied for speech dereverberation since reverberation is considered to be convolutive as well as colored noise. Recently, we revealed that late reverberation is also additive and relatively white interference component in the linear prediction (LP) residual domain. To suppress both late reverberation and additive noise, we have proposed an optimal filter for LP residual estimator (LPRE) based on a constrained minimum mean square error (CMMSE) by using GSVD in single channel speech enhancement, where the algorithm is referred as CMMSE-GSVD-LPRE. Experimental results have shown a better performance of the CMMSE-GSVD-LPRE than spectral subtraction methods, but some residual noise and reverberation components are still audible and annoying. To solve this problem, this paper incorporates the masking properties of the human auditory system in the LP residual domain to further suppress these residual noise and reverberation components while reducing speech distortion at the same time. Various simulation experiments are conducted, and the results show an improved performance of the proposed algorithm. Experimental results with speech recorded in noisy and reverberant environments further confirm the effectiveness of the proposed algorithm in real-world environments.

Bandwidth extension for speech acquired by laser Doppler vibrometer with an auxiliary microphone

This paper proposes to extend the bandwidth of speech acquired by a system combining laser Doppler vibrometer (LDV) and an auxiliary microphone in adverse environments, where this system is referred to as LDV-AM. Traditional bandwidth extension (BWE) algorithms often need two stages to estimate the broadband spectral envelope. In the first stage, the codebook of the broadband spectral envelope is trained by a training data set. In the second stage, the broadband spectral envelope is estimated with the codebook by using different algorithms, such as neural networks and linear mapping. After considering that speech acquired by LDV only contains speech components lower than 2kHz, it seems improper to use the traditional BWE algorithms directly. Because these traditional BWE algorithms are often designed to extend narrowband telephone speech, which is sampled at 8kHz. With the help of the auxiliary microphone, the broadband spectral envelope of speech acquired by LDV can be estimated directly from speech acquired by this microphone. Experimental results show that the proposed BWE algorithm for LDV-AM can greatly improve MOS scores in adverse environments.

Statistical analysis of temporal coherence function and its application in howling detection

Temporal coherence function (TCF) is first defined to measure the correlation between a wave and its delayed version in optics. Now, TCF is a basic physical quantity, that has already been applied to numerous fields. To the best of our knowledge, statistical properties of TCF have not been well studied until now. After defining TCF, this paper gives a new insight into it by studying its statistical properties, where its probability density functions (p.d.f.s) for different types of signals are given. Finally, TCF is applied to detect howling frequency, where we show how to choose the delay time in a best way for practical considerations. There are two benefits by using TCF in howling detection. First, its computation is extremely low. Second, the detection threshold can be easily chosen since TCF is a normalized measurement. Experimental results show the validity of TCF in detecting howling frequency.