Hardware Implementation of Dynamic Synapse Neural Networks for Acoustic Sound Recognition

Identification of acoustic signals in noisy environments remains one of the most difficult of signal processing problems, and is a major obstacle to the high degree of accuracy and speed needed to identify suspicious sounds in high-security, high-safety environments. Given that pattern recognition involves fast identification of a set of fundamental features, the essence of the sound recognition problem is the competing needs for both high-dimensionality representations (necessary for accurate discrimination) and a reasonably limited feature space (necessary for rapid identification). We have previously developed an acoustic recognition capability using a novel biologically based dynamic synapse neural network (DSNN) technology. The DSNN based technology has been demonstrated to classify target sounds with a high degree of accuracy, even in high noise conditions. We have also developed an efficient training algorithm for DSNNs. In this paper we focus on extending the acoustic recognition capability of DSNNs to a hardware implementation and applying the new hardware (acoustic SENTRI) to the problem of gunshot recognition. SENTRI is based on Texas instrument TMS320C6713 DSK and a custom designed data acquisition board. An array of four microphones is attached to the board for sound input. A time delay estimation algorithm (TDE) is employed for triangulation. In field-testing, the SENTRI classifies and localizes over 90% of the trained-for sounds. Performance for firecracker, starter pistol, 9 mm, and 44 caliber, explosion/firing sounds was also tested.