An efficient supervised dictionary learning method for audio signal recognition

Abstract

Machine hearing or listening represents an emerging area. Conventional approaches rely on the design of handcrafted features specialized to a specific audio task and that can hardly generalized to other audio fields. For example, Mel-Frequency Cepstral Coefficients (MFCCs) and its variants were successfully applied to computational auditory scene recognition while Chroma vectors are good at music chord recognition. Unfortunately, these predefined features may be of variable discrimination power while extended to other tasks or even within the same task due to different nature of clips. Motivated by this need of a principled framework across domain applications for machine listening, we propose a generic and data-driven representation learning approach. For this sake, a novel and efficient supervised dictionary learning method is presented.

Real-Time Classification Model of Public Emergencies Using Fusion Expansion Network

“…Its fundamental principle involves the incorporation of update and reset gates to efficiently regulate the information flow. The main purpose of the GRU is to model sequence data, and it is especially good at capturing temporal dependencies in long sequences (Rida et al, 2018). In the field of emergency classification, the GRU performs well in processing this type of data.…”

“…The basic principle is to extract the spatial level features of the input data, layer-by-layer, through convolution operations and pooling operations, thereby achieving effective processing and classification of two-dimensional data such as images. In the field of emergency classification, CNNs have been widely used to process image data and text data (Rida, 2018). Through convolution operations, CNNs are able to capture local features in event data, thereby achieving effective classification of event text and images .…”