An Adaptive Classification Approach for Speech Systems using Tree Structured Vector Quantization

In existing speech coding systems, all Quantizer Code Books are designed to suit the statistical and perceptual characteristics of speech signals of a population of speakers. However, an individual's speech signal does not exhibit, even over a long time, the entire range of characteristics of the population. With the advent of the personal communication systems, personal information might become available and be used to improve the rate - distortion performance of speech coders. In this paper we assess the potential gain of personal speech coding by designing code books for individual speakers. We use a data structure called TREE to implement the vector quantization. This makes the searching time of the speech samples reduce drastically. Vector quantization (VQ) is an appealing coding technique. In tree structured VQ (TSVQ), the key limitation is storage space, since VQ encoding complexity is greatly reduced by the tree structure.