Multimicrophone signal-processing technique to remove room reverberation from speech signals

TLDR

Room reverberation degrades microphone recordings, producing a hollow, echo‑like quality when the microphone is not positioned close to the source. The paper proposes a multimicrophone digital processing scheme to eliminate much of this reverberation distortion. The method splits each microphone signal into frequency bands, co‑phases and sums the band outputs, then adjusts each band’s gain according to the cross‑correlation between corresponding microphone signals. The resulting broadband speech is perceived with markedly reduced reverberation.

Abstract

It is well known that room reverberation can significantly impair one’s perception of sounds recorded by a microphone in that room. Acoustic recordings produced in untreated rooms are characterized by a hollow echolike quality resulting from not locating the microphone close to the source. In this paper we discuss a multimicrophone digital processing scheme for removing much of the degrading distortion. To accomplish this the individual microphone signals are divided into frequency bands whose corresponding outputs are cophased (delay differences are compensated) and added. Then the gain of each resulting band is set based on the cross correlation between corresponding microphone signals in that band. The reconstructed broadband speech is perceived with considerably reduced reverberation.