A Robust Environmental Sound Recognition System using BPNN and RBFNN

Abstract— In a reverberant environment, the performance of acoustic event recognition system can be bolstered by choosing appropriate feature descriptors and classifier techniques. Neural networks are by far providing stunning classification results when compared to other classifiers. This paper analyses two different neural networks and their precision when they both stumble upon same targets in similar environment. The analysis is done on back propagation neural network (BPNN) and radial basis function neural network (RBFNN) with same dataset and then a conclusion is formed on the basis of their performance and efficiency. The experiments on various categories illustrate that the results of recognition for BPNN are significant and effective. KeywordFeature Extraction, Pattern Classification. (Spectral crest, Spectral decrease, Spectral slope, Spectral skewness, Spectral flatness, Back propagation neural network (BPNN), Radial basis function neural network (RBFNN)).