An overview of applications and advancements in automatic sound recognition

Abstract

a b s t r a c tAutomatic sound recognition (ASR) has attracted increased and wide ranging interests in recent years. In this paper, we carry out a review of some important contributions in ASR techniques, mainly over the last one and a half decades. Similar to speech recognition systems, the robustness of an ASR system largely depends on the choice of feature(s) and classifier(s). We take a wider perspective in providing an overview of the features and classifiers used in ASR systems starting from early works in content-based audio classification to more recent developments in applications such as sound event recognition, audio surveillance, and environmental sound recognition. We also review techniques that have been utilized in noise robust sound recognition systems and feature optimization methods. Finally, some of the less commonly known applications of ASR are discussed.

A neural network approach for sound event detection in real life audio

“…A brilliant example of such performance is given in [15], where different ANNs are trained on a big video dataset and then used for different scopes, among which we can also find SED. For a wider overview of the most recent SED techniques the reader can refer to the comprehensive analysis carried out by Sharan et al in [16].…”

Robust sound event classification with bilinear multi-column ELM-AE and two-stage ensemble learning

“…Sound event classification (SEC), also known as acoustic event classification (AEC), which aims to assign a semantic label to the audio content of a short sound chip [1,2], is attracting a growing attention in recent years. SEC has wide potential applications [3], such as acoustic surveillance, bioacoustic monitoring, and environmental sound supervising [1][2][3].…”

“…SEC has wide potential applications [3], such as acoustic surveillance, bioacoustic monitoring, and environmental sound supervising [1][2][3].…”

“…The SEC system usually consists of three components, namely, signal preprocessing, feature extraction, and classification [2]. As a critical factor for the performance of SEC, feature extraction is a challenging task in SEC, and many efforts have been devoted to extract effective feature representation.…”

“…As a critical factor for the performance of SEC, feature extraction is a challenging task in SEC, and many efforts have been devoted to extract effective feature representation. The commonly extracted features for SEC can be roughly divided into the following categories [2]: temporal domain features, frequency domain features, time-frequency image features, cepstral features, modulation frequency features, eigen domain features, and phase space features. However, most of these features are hand-crafted descriptors, which are at a low semantic level, and also generic for different sound datasets without data specificity [4].…”