On the statistical properties of line spectrum pairs
- 19 November 2002
- proceedings article
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 1, 768-771
- https://doi.org/10.1109/icassp.1995.479807
Abstract
In literature, the quantization properties of several representationsof the LPC model have been studied. Until recently, best resultshave generally been obtained with the LSP frequencies. In scalarquantization schemes, the lmmitance Spectrum Pairs (ISP's) [I]perform even slightly better. The good quantization performanceof LSP and ISP can be attributed to their theoretical statisticalproperties: they are uncorrelated when estimated from stationat 3'autoregressive processes, in contrast...Keywords
This publication has 8 references indexed in Scilit:
- Analysis-by-synthesis linear predictive speech coding at 2.4 kbit/sPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2003
- Optimal quantization of LSP parametersIEEE Transactions on Speech and Audio Processing, 1993
- Efficient vector quantization of LPC parameters at 24 bits/frameIEEE Transactions on Speech and Audio Processing, 1993
- Robust and efficient quantization of speech LSP parameters using structured vector quantizersPublished by Institute of Electrical and Electronics Engineers (IEEE) ,1991
- On the statistics of the estimated reflection coefficients of an autoregressive processIEEE Transactions on Acoustics, Speech, and Signal Processing, 1983
- Quantization and bit allocation in speech processingIEEE Transactions on Acoustics, Speech, and Signal Processing, 1976
- Quantization properties of transmission parameters in linear predictive systemsIEEE Transactions on Acoustics, Speech, and Signal Processing, 1975
- Line spectrum representation of linear predictor coefficients of speech signalsThe Journal of the Acoustical Society of America, 1975