An Integrated Real-Time Beamforming and Postfiltering System for Nonstationary Noise Environments

Abstract

<p/> <p>We present a novel approach for real-time multichannel speech enhancement in environments of nonstationary noise and time-varying acoustical transfer functions (ATFs). The proposed system integrates adaptive beamforming, ATF identification, soft signal detection, and multichannel postfiltering. The noise canceller branch of the beamformer and the ATF identification are adaptively updated online, based on hypothesis test results. The noise canceller is updated only during stationary noise frames, and the ATF identification is carried out only when desired source components have been detected. The hypothesis testing is based on the nonstationarity of the signals and the transient power ratio between the beamformer primary output and its reference noise signals. Following the beamforming and the hypothesis testing, estimates for the signal presence probability and for the noise power spectral density are derived. Subsequently, an optimal spectral gain function that minimizes the mean square error of the log-spectral amplitude (LSA) is applied. Experimental results demonstrate the usefulness of the proposed system in nonstationary noise environments.</p

Relative Transfer Function Identification Using Speech Signals

“…In the case of a time-varying system, e.g., moving talkers in hands-free communication scenarios, the proposed method allows to faster and more reliably track the variations. Using the proposed method for the RTF identification, as part of the transfer-function generalized sidelobe canceller (TF-GSC) [2], [4], essentially leads to improved adaptation of the blocking matrix and the noise canceller, and facilitates multichannel signal detection and postfiltering techniques, which employ the transient power ratio between the beamformer output and the reference signals [6], [18], and [19].…”

“…Let , , and be estimates for , , and , respectively. Then (6) where (7) 1 Note that a(t) is generally of infinite length, since it represents the impulse response of the ratio of room transfer functions. However, in real environments, the energy of a(t) often decays exponentially for t > T , where T depends on the reverberation time [2].…”

“…The blocking channel is used for blocking the desired signal and deriving a reference noise signal, and the noise canceller is used for eliminating directional or coherent noise sources. The RTF identification also facilitates multichannel signal detection and postfiltering techniques, which employ the transient power ratio between the beamformer output and the reference signals [5], [6].…”

Real-Time Dual-Microphone Speech Enhancement

“…To discriminate desired speech transients from interfering transients, he used both the GSC beamformer primary output and the reference noise signals. To get a real-time implementation of the method, the author suggested in an earlier paper (Cohen, 2003a), feeding back to the beamformer the discrimination decisions made by the postfilter.…”

Speech Enhancement Using Gaussian Scale Mixture Models

“…Algorithms based on multiple microphones [2]- [4] and single microphone have also been successful in achieving some measure of speech enhancement [5]- [13].…”