Detecting when voice is or is not present is an outstanding problem for speech transmission, enhancement and recognition. Here we present a novel multichannel source activity detector that exploits the spatial localization of the target audio source. The detector uses an array signal processing technique to maximize the signal-to-interference ratio for the target source thus decreasing the activity detection error rate. We compare our two-channel voice activity detector (VAD) with the AMR voice detection algorithms on real data recorded in a noisy car environment. The new algorithm shows improvements in error rates of 55-70% compared to the state-of-the-art adaptive multi-rate algorithm AMR2 used in present voice transmission technology.