"polyaural" array processing for automatic speech recognition in degraded environments.

摘要: In this paper we present a new method of signal processing for robust speech recognition using multiple microphones. The method, loosely based on the human binaural hearing system, consists passing signals detected by microphones through bandpass filtering and nonlinear halfwave rectification operations, then cross-correlating outputs from each channel within frequency band. These operations provide rejection off-axis interfering signals. are repeated (in non-physiological fashion) negative signal, an estimate desired is obtained combining positive outputs. We demonstrate that use approach provides substantially better accuracy than delay-and-sum beamforming same sensors target in presence additive broadband maskers. Improvements reverberant environments tangible but more modest.