Home > Research > Research Areas  
Three-dimensional sound source localization using 2-channel rotating microphone array
 
The three-dimensional (3-D) sound source localization based on measured acoustic signals is one of the key techniques for human-robot interactions, monitoring systems, tele-conference for remote users. Especially, the three-dimensional localization ability, which is able to estimate source¡¯s azimuth and elevation angles at the same time, is demanded by the mobile platform applications such as intelligent robots.

This research addresses the (3-D) localization problem using 2-channel rotating microphone array, which is supposed to be installed on mobile platform. In the perspectives to decouple 3-D localization into two disjoint estimation problem: azimuth and elevation angle estimations, the (source) direction- and (shift) angle-dependent inter-channel time difference (ICTD) trajectory is proposed as a new localization cue and realized by 2-channel rotating microphone array. The mathematical formula of this localization cue shows that the ICTD trajectory¡¯s mean and amount of its phase shift are strongly related to the azimuth and elevation angles separately in a form of one-to-one relation. Besides, the cone of confusion problem that has not been solved by the previous localization methods is also treated well. In addition, the probabilistic approach for 3-D localization is investigated on the basis of the Bayesian inference. By the simulation results, it is found that the localization performance is strongly dependent on the azimuth angle only as expected, and this performance is quite reasonably interpretable by the trajectory¡¯s amplitude. The two experiments were conducted in the room environment for testing its feasibility. The localization performance of sources in the median plane is fairly good (mean error: 3.23o, standard deviation: 1.33o). In the other experiment of involving a voice signal, it is found that even the partially valuable ICTD trajectory, which is caused by low signal to noise ratio (SNR) condition, can be also efficiently applicable to 3-D localization. Thus, it is concluded that to achieve 3-D localization ability using only two microphones, the proposed ICTD trajectory is quite useful for 3-D sound source localization.

Figure. 2-channel rotating microphone array