Machine Listening in Multisource Environments (CHiME) 2011
This paper describes a novel adaptive beamforming technique, for speech enhancement applications, designed to be robust to nonstationary interfering sources in noisy and reverberant environments. The proposed beamforming architecture aims at extracting the desired source signal and suppressing interfering signals in a multisource environment with unknown a priori conditions. This purpose is realized by means of a multi-stage collaborative generalized sidelobe canceller. The trademark of this architecture relies on a two-level convex combination of two multiple-input single-output (MISO) adaptive systems, which improves the beamformer capability to track undesired sources, in order to achieve a stronger suppression of interfering signals. The potency of the proposed architecture is proved enhancing the speech quality of the desired source in a handsfree teleconferencing application.
Index Terms. Nonstationary Adaptive Beamforming, Speech Enhancement, Combination of Adaptive Filters
Bibliographic reference. Comminiello, Danilo / Scarpiniti, Michele / Parisi, Raffaele / Cirillo, Albenzio / Falcone, Mauro / Uncini, Aurelio (2011): "Multi-stage collaborative microphone array beamforming in presence of nonstationary interfering signals", In CHiME-2011, 64-67.