Multi Target Acoustic Source Tracking using Track Before Detect

Particle Filter-based Source Localisation algorithms attempt to track the position of a sound source - a person speaking in a room - based on the current data from a distributed microphone array as well as all previous data up to that point. This paper introduces a multi-target methodology for acoustic source tracking. The methodology is based upon the Track Before Detect (TBD) framework. The algorithm also implicitly evaluates the source activity using a variable appended to the state vector. Examples of typical tracking performance are given using a set of real speech recordings with two sources active simultaneously.