In this paper, we present a hybrid time-frequency approach for the detection of audio signal patterns for the purpose of surveillance applications. The proposed detection algorithm evolves through two main processing phases, denoted as coarse and fine, respectively. The key idea is that of distinguishing classes of audio signals on the basis of spectral “signatures.” However, we use different processing phases, in the time (“coarse”) and frequency (“fine”) domains, in order to reduce the computational complexity. In fact, fine processing (in the frequency domain) is carried out only when an "atypical" audio signal is detected, on the basis of low-complexity energy detection, with coarse processing. The performance of the proposed detection algorithm is investigated also using experimental audio signals acquired with the microphone module of a multi-sensorial wireless sensor network (WSN) node prototype. The obtained results show that the proposed approach allows to effectively detect a signal of interest, thus making it suitable to surveillance applications and in general for distributed sensing applications.
A hybrid time-frequency audio signal pattern detection algorithm for surveillance applications
MARTALO', MARCO;
2013-01-01
Abstract
In this paper, we present a hybrid time-frequency approach for the detection of audio signal patterns for the purpose of surveillance applications. The proposed detection algorithm evolves through two main processing phases, denoted as coarse and fine, respectively. The key idea is that of distinguishing classes of audio signals on the basis of spectral “signatures.” However, we use different processing phases, in the time (“coarse”) and frequency (“fine”) domains, in order to reduce the computational complexity. In fact, fine processing (in the frequency domain) is carried out only when an "atypical" audio signal is detected, on the basis of low-complexity energy detection, with coarse processing. The performance of the proposed detection algorithm is investigated also using experimental audio signals acquired with the microphone module of a multi-sensorial wireless sensor network (WSN) node prototype. The obtained results show that the proposed approach allows to effectively detect a signal of interest, thus making it suitable to surveillance applications and in general for distributed sensing applications.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
