Security monitoring based on joint automatic speaker recognition and blind source separation

The aim of this paper is to introduce an enhanced approach for standard Automatic Speaker Recognition (ASR) systems in noisy environment in conjunction with a Blind Source Separation (BSS) algorithm. This latter is able to discern between interfering noise signals and the reference speech signal, hence it can be consider as a necessary preprocessing step. The main problem of the proposed approach lies in the not removable ambiguities typically of the BSS algorithms. In order to overcome to this drawback, a geometrical constraint is also added to the learning algorithm. A practical example shows the effectiveness of the proposed approach in terms of recognition accuracy.