Real-Time Implementation of a Particle Filter with Integrated Voice Activity Detector for Acoustic Speaker Tracking

In noisy and reverberant environments, the problem of acoustic source localisation and tracking (ASLT) using an array of microphones presents a number of challenging difficulties. One of the main issues when considering real-world situations involving human speakers is the temporally discontinuous nature of speech signals: the presence of silence gaps in the speech can easily misguide the tracking algorithm, even in practical environments with low to moderate noise and reverberation levels. This work focuses on a realtime implementation of the ASLT algorithm, which circumvents this problem by integrating measurements from a voice activity detector (VAD) within the tracking algorithm framework. The algorithm is here optimized for low computational complexity, and is implemented on a PC based real-time system. The resulting computational load is calculated and is presented along with real measurements of the true execution speed for the considered algorithm implementation. The results show that the algorithm is suitable for implementation in currently existing low-power embedded systems