A hybrid noise canceller with a real-time adaptive Wiener filter and a geometric-based voice-activity detector for an automotive application

In real‐time environments a speech recognition system in a car has to receive the driver's voice only while suppressing the background noise. This paper presents a hybrid real‐time adaptive filter that operates within a geometrical zone defined around the head of the desired speaker. Any sound outside of this zone is considered to be noise and is suppressed. As this defined geometrical zone is small, it is assumed that only driver's speech is incoming from this zone. The technique uses three microphones to define a geometric‐based voice–activity detector (VAD) to cancel the unwanted speech coming from outside of the zone. However, when unwanted speech and desired speech are incoming at the same time, the VAD fails to identify the unwanted speech or desired speech. In such a situation an adaptive Wiener filter is switched on for noise reduction. In the case of sole unwanted speech incoming from outside of a desired zone, this speech is muted at the output of the hybrid noise canceller. In the case of desired and unwanted speeches incoming together, the SNR is improved by as much as 20 dB. Copyright © 2009 John Wiley & Sons, Ltd.