Voice Signal Compression and Spectrum Analysis (VSCSA): Suitable For Pervasive Computing and Limited Storage Devices Using MatLab

Abstract

In this paper, Voice Signal Compression and Spectrum Analysis (VSCSA) is a technique that is used to compress transparent high quality voice signals upto 45% - 60% of the source file at low bit rate 45kbps with same extension (i.e. .wav to .wav), then voice spectrum analysis (VSA) is started. Voice Signal Compression (VSC) is done by using adaptive wavelet packet a tool of MatLab for decomposition and psychoacoustic model implementation. Entropy & Signal to Noise Ratio (SNR) of given input voice signal is computed during VSC. Filter-bank is used according to psychoacoustic model criteria and computational complexity of the decoder for VSC. Bit allocation method is used that also take input from the psychoacoustic model. The purpose of VSCSA is to compress the voice signals with same extension with the help of VSC and then distinguish between constitutional and unconstitutional voice with the help of VSA according to various parameters of DSP. If a voice signal is compressed first then spectrum analysis will be very fast because selected .wav file will take very short time for execution of various DSP parameters that gives better result. For example, if a device is bolted with DSP parameters then it can unbolt only when bolted device is recognized same DSP parameters from the .wav warehouse. This work is suitable for pervasive computing, Internet, and limited storage devices because of reduction in file size and fast execution.