Tree structured vector quantization based technique for speech compression

Abstract

In this paper, the Tree-Structured Vector Quantization (TSVQ) method for proficient speech compression is presented. Efficient utilization of memory is always needed when analog-encoded or digitized data such as image, audio, videos, portable files are need to store and/or convey to digital channels. Compression offers betterments with storage requirements while transmitting the encoded signals with lossy and lossless techniques. Lossy compression is always intended for compression of high volume data with Scalar Quantization (SQ) and Vector Quantization (VQ). The Tree based VQ method is used with hieratically organized binary sequences of codeword of data (speech) for compression with reduced and minimized arithmetic calculation requirements. Speech compression has been gained by compressed-codebook coefficients and structured in binary fashion. The quantization noise ratio with signal power is obtained efficiently around less than 1.082 dB. Shared codebook method described in this TSVQ algorithm achieves 3.6 reduced storage requirements of factor 5 to 3.