Spectral vs. spectro-temporal features for acoustic event detection

Abstract

Automatic detection of different types of acoustic events is an interesting problem in soundtrack processing. Typical approaches to the problem use short-term spectral features to describe the audio signal, with additional modeling on top to take temporal context into account. We propose an approach to detecting and modeling acoustic events that directly describes temporal context, using convolutive non-negative matrix factorization (NMF). NMF is useful for finding parts-based decompositions of data; here it is used to discover a set of spectro-temporal patch bases that best describe the data, with the patches corresponding to event-like structures. We derive features from the activations of these patch bases, and perform event detection on a database consisting of 16 classes of meeting-room acoustic events. We compare our approach with a baseline using standard short-term mel frequency cepstral coefficient (MFCC) features. We demonstrate that the event-based system is more robust in the presence of added noise than the MFCC-based system, and that a combination of the two systems performs even better than either individually.

Optimizing acoustic feature extractor for anomalous sound detection based on Neyman-Pearson lemma

“…In addition, F and G were implemented using fully-connected DNNs, and the symmetric network architecture of F was used for that of G. Then, F and G were trained to maximize (12) and to minimize (19) and (20), alternately.…”

“…To extract a set of acoustic features for the soundidentification problem, feature-extractor-optimization methods have been actively investigated [12], [13], [14]. These studies have revealed that it is necessary to determine both spectral and temporal characteristics to accurately identify various sounds.…”

Acoustic feature extraction by statistics based local binary pattern for environmental sound classification

“…Recently, the method of matching pursuit that sparsely decomposes the signal by using over-complete dictionaries has been successfully applied to classify the environmental sounds [2,5]. As in the matching pursuit, the non-negative matrix factorization (NMF) also works on sparse factorization of signals with learning the dictionary; Cotton and Ellis [8] employ the NMF to construct acoustic event-based patch features from a spectrogram. Ye et al [3] utilize the acoustic subspace extracted from sound clips in the kernel-based framework.…”

On the Joint Use of NMF and Classification for Overlapping Acoustic Event Detection

“…NMF-related methods can be separated in those that exploit the NMF activations directly to perform event detection [8,11], and in those that employ a classifier trained on these activations [12,13]. Based on the fact that NMF-based approaches can benefit from the creation of a Mixture of Local Dictionaries (MLD) [14], in [15] the authors propose a classifier-based NMF system using MLDs for improved detection performance.…”