Efficiency of Ordered Codebook Learning VectorQuantization for Speech Compression

Abstract

Combined compression and classification problems are becoming increasingly important in many applications with large amounts of data and large sets of classes. This article presents the efficiency of ordered codebook learning vector quantization (OC-LVQ) for speech compression. The algorithm is based on competitive networks. It is developed and analyzed a learning vector quantization based algorithm for combined speech compression and classification. The Peak Signal to Noise Ratio (PSNR), Signal to Noise Ratio (SNR), and Normalized Root Mean Square Error (NEMSE) are used to measure the quality of speech signal. It provides the maximum quality at 28.9432 dB and 15.0333 dB for SNR and PSNR respectively. Also the minimum error of NEMSE is 0.1578. Opinion Score (MOS). The results show that the DWT achieves greater performance than other two techniques employed in this research. This article presents an exploitation of Learning Vector Quantization (LVQ), in ordered codebook for speech compression. The adopted techniques are evaluated based on Signal to Noise Ratio (SNR), Peak Signal to Noise Ratio (PSNR), Normalized Root Mean Square Error (NRMSE). It is organized as follows. Section II describes the linear predictive coefficients while Section III details the LVQ neural network. Section IV details experiment while V shows its simulations results. Finally, Section VI concludes this work.