In this work a challenging scenario concerning hands-free continuous speech recognition is investigated. A set of experiments was carried out using microphone arrays having different numbers of omnidirectional sensors and that were placed at different angles and distances from the talker. Both real and simulated array signals, obtained by means of the image method, were used. An enhanced input to a recognizer based on Hidden Markov Models was obtained by a time delay compensation module providing a beamformed signal. HMM adaptation was used to improve recognition performance in the various acoustic conditions.