Binaural noise reduction via cue-preserving MMSE filter and adaptive-blocking-based noise PSD estimation

Abstract

Binaural noise reduction, with applications for instance in hearing aids, has been a very significant challenge. This task relates to the optimal utilization of the available microphone signals for the estimation of the ambient noise characteristics and for the optimal filtering algorithm to separate the desired speech from the noise. The additional requirements of low computational complexity and low latency further complicate the design. A particular challenge results from the desired reconstruction of binaural speech input with spatial cue preservation. The latter essentially diminishes the utility of multiple-input/single-output filter-and-sum techniques such as beamforming. In this paper, we propose a comprehensive and effective signal processing configuration with which most of the aforementioned criteria can be met suitably. This relates especially to the requirement of efficient online adaptive processing for noise estimation and optimal filtering while preserving the binaural cues. Regarding noise estimation, we consider three different architectures: interaural (ITF), cross-relation (CR), and principal-component (PCA) target blocking. An objective comparison with two other noise PSD estimation algorithms demonstrates the superiority of the blocking-based noise estimators, especially the CR-based and ITF-based blocking architectures. Moreover, we present a new noise reduction filter based on minimum mean-square error (MMSE), which belongs to the class of common gain filters, hence being rigorous in terms of spatial cue preservation but also efficient and competitive for the acoustic noise reduction task. A formal real-time subjective listening test procedure is also developed in this paper. The proposed listening test enables a real-time assessment of the proposed computationally efficient noise reduction algorithms in a realistic acoustic environment, e.g., considering time-varying room impulse responses and the Lombard effect. The listening test outcome reveals that the signals processed by the blocking-based algorithms are significantly preferred over the noisy signal in terms of instantaneous noise attenuation. Furthermore, the listening test data analysis confirms the conclusions drawn based on the objective evaluation.

Cue-Preserving MMSE Filter with Bayesian SNR Marginalization for Binaural Speech Enhancement

“…The optimal MMSE filter in (3) coincides with the cue-preserving MMSE filter in [16], which has proven itself to well preserve binaural cues. Even though the authors presented a variety of a-priori SNR estimators the listening experience indicates a requirement for further improvement.…”

“…where the spectral density of left and right noise are assumed to be equal Φn l n l = Φn r nr = Φnn as in [16]. We then introduce a new apriori SNR definition η = (||H|| 2 Φs)/Φnn, including the acoustic channel information, here the head-related transfer function (HRTF) H = [H l Hr] T , to express the MMSE filter in a univariate form…”

“…Integration in D once again resolves into 1 for the minuend and invokes the MWF for the subtrahend as in (3). Equating (16) to zero and rearranging we obtain…”

“…They can be expressed, e.g., by the minimum, maximum, and average of two independent single-channel gain functions [14]. A joint cost function for signal enhancement at the left and right ear has eliminated a need for heuristic combination of bilateral filters for cuepreserving spectral enhancement [16]. Claims of optimality are expressed with methods of spectral-distance minimization [17].…”

Real-Time Binaural Speech Separation with Preserved Spatial Cues

“…The second category calculates a real-valued spectro-temporal mask and then applies the same mask to both left and right microphone channels [6,[8][9][10][11][12]. Because both sides obtain the same zerophase gain, the original interaural cues are preserved.…”