Switched Multistage Vector Quantizer

This paper investigates the use of a new hybrid vector quantizer called Switched Multi stage vector quantization (SWMSVQ) technique using hard and soft decision schemes, for coding of narrow band speech signals. This technique is a hybrid of Switch vector quantization technique and Multi stage vector quantization technique. SWMSVQ quantizes the linear predictive coefficients (LPC) in terms of the line spectral frequencies (LSF). The spectral distortion performance, computational complexity and memory requirements of SWMSVQ using hard and soft decision schemes are compared with Split vector quantization (SVQ) technique, Multi stage vector quantization (MSVQ) technique, Switched Split vector quantization (SSVQ) technique using hard decision scheme, and Multi Switched Split Vector quantization (MSSVQ) technique using hard decision scheme. From results it is proved that SWMSVQ using soft decision scheme is having less spectral distortion, computational complexity and memory requirements when compared to SVQ, MSVQ, SSVQ and SWMSVQ using hard decision scheme, but high when compared to MSSVQ using hard decision scheme. So from results it is proved that SWMSVQ using soft decision scheme is better when compared to SVQ, MSVQ, SSVQ and SWMSVQ using hard decision schemes in terms of spectral distortion, computational complexity and memory requirements but is having greater spectral distortion, computational complexity and memory requirements when compared to MSSVQ using hard decision.