Multichannel Post-Filtering in Nonstationary Noise Environments

Abstract

Abstract-In this paper, we present a multichannel post-filtering approach for minimizing the log-spectral amplitude distortion in nonstationary noise environments. The beamformer is realistically assumed to have a steering error, a blocking matrix that is unable to block all of the desired signal components, and a noise canceller that is adapted to the pseudo-stationary noise but not modified during transient interferences. A mild assumption is made that a desired signal component is stronger at the beamformer output than at any reference noise signal, and a noise component is strongest at one of the reference signals. The ratio between the transient power at the beamformer output and the transient power at the reference noise signals is used to indicate whether such a transient is desired or interfering. Based on a Gaussian statistical model and combined with an appropriate spectral enhancement technique, we derive estimators for the signal presence probability, the noise power spectral density, and the clean signal. The proposed method is tested in various nonstationary noise environments. Compared with single-channel post-filtering, a significantly reduced level of nonstationary noise is achieved without further distorting the desired signal components.Index Terms-Acoustic noise measurement, adaptive signal processing, array signal processing, signal detection, spectral analysis, speech enhancement.

Speech Enhancement via Combination of Wiener Filter and Blind Source Separation

“…Beamforming [6,7], post-filtering [4,5] and multi-channel Wiener filtering [3] are often used in multi-channel noise reduction. However, beamforming [6,7] requires a robust estimation of the direction of the target speech.…”

“…In the given sound material that was recorded in highly reverberated environment (e.g., 300 ms reverberation), this was difficult to estimate correctly and therefore we decided against using beamforming. Post-filtering [4,5] on the other hand, assumes that background noises in each channel are only weakly correlated and not directional. In the given sound material noise sources were often directional, and therefore post-filtering was not applied.…”

“…Traditional speech enhancement algorithms often work only in narrowly specified conditions or with specific noise statistics. Algorithms exist and work well, when the background noise is stationary and non-speech [2][3][4][5], however, these algorithms often fail when competing speakers are present. A possible solution to this problem is to use source separation algorithm like beamform-ing [6,7].…”

A generalized estimation approach for linear and nonlinear microphone array post-filters

“…Since, frames with SNRs above 35dB do not contribute significantly to the overall speech quality and frames consisting of silence can have SNRs with extreme negative values, that do not reflect the perceptual contribution of the signal, the SNR at each frame is limited to the range of (−10, 35dB). To assess the speech quality of the enhanced output signal we use the Log-Area-Ratio distance (LAR), the Log-Likelihood Ratio (LLR), the Itakura-Saito Distortion (IS) [27] and the Log-Spectral Distance (LSD) [12]. These measures are found to have a high correlation with the human perception.…”

“…This is due to the requirement for knowledge of second order statistics for both the signal and the corrupting noise that makes these filters signal-dependent. A variety of post-filtering techniques trying to address this issue have been proposed in the literature [7][8][9][10][11][12]. A quite common method for the formulation of the post-filter transfer function is based on the use of the auto-and cross-power spectral densities of the multichannel input signals [2,7,11].…”