An auditory classifier employing a wavelet neural network implemented in a digital design

This work explores the use of a wavelet transform, a feature extractor mechanism, and a neural network to classify audio samples as belonging to either a voice class, or a music class. The proposed system was implemented in a digital design using VHDL and was synthesized with the Synopsys Design Compiler, using the LSI-10 K synthesized library cells with a clock frequency of 11.025 kHz. This design of a wavelet neural network was effective in correctly identifying the test data sets.