Multi-modal Fusion Framework with Particle Filter for Speaker Tracking

In the domain of Human-Computer Interaction (HCI), the main focus of the computer is to interpret the external stimuli provided by users. Moreover in the multi-person scenarios, it is important to localize and track the speaker. To solve this issue, we introduce here a framework by which multi-modal sensory data can be efficiently and meaningfully combined in the application of speaker tracking. This framework fuses together four different observation types taken from multi-modal sensors. The advantages of this fusion are that weak sensory data from either modality can be reinforced, and the presence of noise can be reduced. We propose a method of combining these modalities by employing a particle filter. This method offers satisfied real-time performance. We demonstrate results of a speaker localization in twoand three-person scenarios.