Acoustic novelty detection based on AHLAC and NMF

The importance of video surveillance applications has been increasing with the increase of crime and terrorism. In addition to traditional video cameras, the use of acoustic sensors in surveillance and monitoring applications is also becoming increasingly important. In this paper, we apply High-Order Local Auto-Correlation (HLAC), which has succeeded in video surveillance applications, to extract features from acoustic signals in order to construct an acoustic surveillance system based on a novelty detection approach. We also apply Non-Negative Matrix Factorization (NMF) to this problem. Experiment results confirmed that the AHLAC-based method outperforms the NMF-based and the cepstrum-based methods under all SNR conditions. The combined NMF-AHLAC method was able to improve the Equal Error Rate (EER) at lower SNRs, although the EERs at higher SNRs tend to degrade.