《A Dual-Microphone Speech Enhancement Algorithm Based on the Coherence Function》.pdf

IEEE TRANSACTIONS ON AUDIO, SPEECH, AND LANGUAGE PROCESSING, VOL. 20, NO. 2, FEBRUARY 2012 599 A Dual-Microphone Speech Enhancement Algorithm Based on the Coherence Function Nima Yousefian and Philipos C. Loizou, Senior Member, IEEE Abstract—A novel dual-microphone speech enhancement tech- by a single well-defined directional noise source and character- nique is proposed in the present paper. The technique utilizes the ized by high correlation between noise signals; and 3) diffuse coherence between the target and noise signals as a criterion for noise, which is characterized by uncorrelated noise signals of noise reduction and can be generally applied to arrays with closely spaced microphones, where noise captured by the sensors is highly equal power propagating in all directions simultaneously [3]. correlated. The proposed algorithm is simple to implement and Performance of speech enhancement methods is strongly depen- requires no estimation of noise statistics. In addition, it offers the dent on the characteristics of the environmental noise they are capability of coping with multiple interfering sources that might tested in. Hence, the performance of methods such as [4], [5] be located at different azimuths. The proposed algorithm was that work well in the diffuse field, starts to degrade when tested evaluated with normal hearing listeners using intelligibility lis- in coherent noise fields. tening tests and compared against a well-established beamforming algorithm. Results indicated large gains in speech intelligibility Traditionally, only one microphone is used in speech en- relative to the baseline (front