A Target Guided Subband Filter for Acoustic Event Detection in Noisy Environments Using Wavelet Packets

Abstract

This paper deals with acoustic event detection (AED), such as screams, gunshots and explosions, in noisy environments. The main aim is to improve the detection performance under adverse conditions with a very low signal-to-noise ratio (SNR). A novel filtering method combined with an energy detector is presented. The wavelet packet transform (WPT) is first used for time-frequency representation of the acoustic signal. The proposed filter in the wavelet packet domain then uses a priori knowledge of the target event and an estimate of noise features to selectively suppress the background noise. It is in fact a content-aware band-pass filter which can automatically pass the frequency bands that are more significant in the target than in the noise. Theoretical analysis shows that the proposed filtering method is capable of enhancing the target content while suppressing the background noise, for signals with a low SNR. A condition to increase the probability of correct detection is also obtained. Experiments have been carried out on a large dataset of acoustic events that are contaminated by different types of environmental noise and white noise with varying SNRs. Results show that the proposed method is more robust and better adapted to noise than ordinary energy detectors and it can work even with an SNR as low as -15dB. A practical system for real time processing and multi-target detection is also proposed in this work.

Adaptive Noise Reduction for Sound Event Detection Using Subband-Weighted NMF

“…Two weighting strategies are derived, namely: frequency-based and temporal-based. The first one is based on subband importance [26], and is used to enhance those dominant frequencies of the event class, and the latter is used to re-weight different frames and to highlight those frames with a higher event presence probability. A noteworthy property of the proposed weighting scheme is its adaptive noise reduction ability.…”

Random forest classification based acoustic event detection utilizing contextual-information and bottleneck features

“…The works in [22,23,24,25,26] used short frame based acoustic features. Moreover, these frame based acoustic features are extracted without any prior knowledge of the target events, which has been shown to be useful in [27,28]. In [20], a 100ms long window was used while extracting the acoustic features and significant improvements have been achieved.…”

Design of an Acoustic Target Intrusion Detection System Based on Small-Aperture Microphone Array

“…Hence, effective intrusion detection is a great challenge because of the processing capability and battery life resource limitations of sensor nodes that make up most economically viable WSNs [2]. Acoustic sensor has the advantages of low cost, low power consumption, and non-line of sight measurement which is widely adopted for a long time military terrain reconnaissance [3,4,5,6,7,8,9]. …”

“…[4]. Therefore, features employed by researchers are normally extracted from frequency domain [3,5] and time-frequency domain [6]. The harmonic line association (HLA) feature was utilized in [7,8] while it is difficult to extract HLA feature due to the non-stationary and wideband characteristic of vehicle acoustic signals.…”