Vector-Predictive Speech Coding with Quantization Noise Modelling

  • Søren Vang Andersen
  • Søren Holdt Jensen
  • Egon Hansen
Part of the Applied and Numerical Harmonic Analysis book series (ANHA)


This chapter studies the modelling of quantization noise in a predictive coding algorithm. The specific algorithm is a low-delay, vector-predictive transform coder for speech signals. The coder uses a Kaiman filter with a backward estimated LPC model for the speech signal combined with an additive noise model for the quantization noise. Three different approaches to the quantization noise problem are described and compared. These are: neglecting the quantization noise in the prediction, modelling it as an uncorrelated additive noise source, and modelling it as a noise source correlated with the quantizer input. Simulations indicate that this last approach is the most efficient among those described. Moreover, we study the choice of measurements for the Kaiman filter in connection with signal smoothing. In the coding algorithm described, the measurements are obtained by a transform matrix. Simulations indicate that an overlap in the vectors, which are transformed, provides an improvement in coder performance.


Kalman Filter Speech Signal Quantization Noise Speech Code Kalman Gain 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [1]
    S.V. Andersen, S.H. Jensen, and E. Hansen. Quantization Noise Modeling in Low-Delay Speech Coding. In IEEE Workshop on Speech Coding for Telecommunications Proceedings, pp. 65-66, Pocono Manor, PA, 1997.Google Scholar
  2. [2]
    S.V. Andersen, S.H. Jensen, and E. Hansen. Quantization Noise Modeling in Low-Delay Vector-Predictive Speech Coding. In Signal Analysis and Prediction I, pp. 299–302. ICT Press Prague, Czech Republic, 1997.Google Scholar
  3. [3]
    S.V. Andersen, M. Olesen, S.H. Jensen, and E. Hansen. Innovation Coding with a Cross-Correlated Quantization Noise Model. In Signal Processing VIII: Theories and Applications, pp. 1019–1022. Edizioni LINT Trieste, Italy, 1996.Google Scholar
  4. [4]
    P.J. Brockwell and R.A. Davis. Time Series: Theory and Methods. Springer-Verlag, 1991.Google Scholar
  5. [5]
    S. Crisafulli. Adaptive Speech Coding via Feedback Techniques. Ph.D. Thesis, Department of Systems Engineering, Research School of Physical Sciences and Engineering, The Australian National University, Canberra, Australia, 1992.Google Scholar
  6. [6]
    S. Crisafulli, J.D. Mills, and R.R. Bitmead. Kaiman Filtering Techniques in Speech Coding. In Proc. IEEE Int. Conf. Acoust., Speech, Signal Processing, 1, pp. 77–80, San Francisco, California, 1992.Google Scholar
  7. [7]
    V. Cuperman and A. Gersho. Vector-Predictive Coding of Speech at 16 kbits/s. IEEE Trans. Comm., 33:685–696, 1985.CrossRefGoogle Scholar
  8. [8]
    A. Gersho and R.M. Gray. Vector Quantization and Signal Compression. Kluwer Academic Publishers, 1992.Google Scholar
  9. [9]
    J.D. Gibson. Optimal and Suboptimal Estimation in Differential PCM and Adaptive Coding Systems. In Proc. IEEE Conf. Decision and Control, p. 353, Houston, TX, 1975.Google Scholar
  10. [10]
    J.D. Gibson. Sequential Filtering of Quantization Noise in Differential Encoding Systems. In Proc. Int. Conf. Cybernetics and Society, pp. 685-689, Washington, DC, 1976.Google Scholar
  11. [11]
    J.D. Gibson. Quantization Noise Filtering in ADPCM Systems. IEEE Trans. Sys., Man, and Cybern., 10:529–536, 1980.CrossRefGoogle Scholar
  12. [12]
    J.E. Gunn and A.P. Sage. Speech Data Rate Reduction I: Applicability of Modern Estimation Theory. IEEE Trans. Aerospace Electron. Syst., 9:130–138, 1973.CrossRefGoogle Scholar
  13. [13]
    S. Haykin. Adaptive Filter Theory, 3rd edition. Prentice Hall, Engelwood Cliffs, NJ, 1996.Google Scholar
  14. [14]
    ITU International Telecommunication Union, Geneva. Coding of Speech at 16 kbit/s Using Low-Delay Code Excited Linear Prediction, Recommendation G.728, 1992.Google Scholar
  15. [15]
    N.S. Jayant. Adaptive Quantization With a One-Word Memory. The Bell System Technical Journal, 52:1119–1144, 1973.CrossRefGoogle Scholar
  16. [16]
    R.E. Kaiman. A New Approach to Linear Filtering and Prediction Problems. Trans. ASME, J. Basic Engineering, 82:35–45, 1960.CrossRefGoogle Scholar
  17. [17]
    J.L. Melsa and R.B. Kolstad. Kaiman Filtering of Quantization Error in Digitally Processed Speech. In Conf. Rec. IEEE Int. Conf. Commun., pp. 310-313, Chicago, Illinois, 1977.Google Scholar
  18. [18]
    G. Pirani and C. Scagliola. Performance Analysis of DPCM Speech Transmission Systems using the Kaiman Predictors. In Conf. Rec. 1976 IEEE Int. Conf. on Acoustics, Speech and Signal Processing, pp. 262-265, Philadelphia, Pennsylvania, 1976.Google Scholar
  19. [19]
    T.V. Ramabadran and D. Sinha. On the Selection of Measurements in Least-Squares Estimation. In Proc. IEEE Int. Conf. on Systems Eng., pp. 221-226, Dayton, Ohio, 1989.Google Scholar
  20. [20]
    T.V. Ramabadran and D. Sinha. Speech Data Compression Through Sparse Coding of Innovations. IEEE Trans. Speech, Audio Processing, 2:274–284, 1994.CrossRefGoogle Scholar
  21. [21]
    T.V. Ramabadran and D. Sinha. Speech Coding using Least-Squares Estimation. In B.S. Atal, V. Cuperman, and A. Gersho, eds, Advances in Speech Coding, chapter 33, pp. 349-359, Kluwer, 1991.Google Scholar
  22. [22]
    C.R. Watkins, S. Crisafulli, and R.R. Bitmead. Reduced Complexity Kaiman Filtering for Signal Coding. In Int. Workshop on Intelligent Signal Processing and Communication Systems, pp. 269-273, Sendai, Japan, 1993.Google Scholar

Copyright information

© Springer Science+Business Media New York 1998

Authors and Affiliations

  • Søren Vang Andersen
    • 1
  • Søren Holdt Jensen
    • 1
  • Egon Hansen
    • 1
  1. 1.Institute of Electronic Systems, Center for Personkommunikation (CPK)Aalborg UniversityAalborgDenmark

Personalised recommendations