This paper proposes a new method for estimating the talker's head orientation in a smart room equipped with microphone arrays. The acoustic processing is based on the use of a coherence measure derived from the Cross-power spectrum phase analysis, commonly used for speaker localization and tracking purposes. An Oriented Global Coherence Field function is then introduced to assign to a given point in space different scores according to the possible orientation of the acoustic source.

Preliminary experiments were conducted in the CHIL smart room available at ITC-irst laboratories. A small database was collected, given one speaker uttering a given sentence in different directions. Although the database is rather small, results show that the proposed method is promising and ensures a good estimation of the head orientation.