Application of sorted codebook vector quantization to spectral coding of speech

Abstract

A new vector quantization method, namely sorted codebook vector quantization (SCVQ) is presented in this article. The paper explains the principles of this method, including training and optimization of the associated codebook. It is shown that this quantizer can be implemented efficiently with almost similar computational complexity to tree-searched vector quantization (TSVQ) and the storage cost of that is the same as unstructured VQ (i.e less than TSVQ). Application of SCVQ to quantization of Line Spectral Frequencies (LSFs), which are the most popular parameters for spectrum quantization in speech coders using linear prediction model, is described. Superior performance of the new method is verified through experimental simulations.

Related Papers

• → Applying LVQ Techniques to Compress Historical Information in Sensor Networks(2005)2 cited
• A New Initial Codebook Algorithm of Learning Vector Quantization(2006)
• → A New Initial Codebook Algorithm for Learning Vector Quantization(2007)1 cited
• → Ordered vector quantization for neural network pattern classification(2002)1 cited
• A Speech Wave Coding Based on Modified Vector Quantization Technique(2007)