1. Figure 11.1 Linear system model for a signal s[n].

2. Figure 11.2 Inverse filter formulation for all-pole signal modeling.

3. Figure 11.3 Linear prediction formulation for all-pole signal modeling.

4. Figure 11.4 Linear system model for a random signal s[n].

5. Figure 11.5 Examples of deterministic and random outputs of a 1st-order all-pole system.

6. Figure Illustration (for p = 5) of computation of prediction error for the autocorrelation method. (Square dots denote samples of hA[n−m] and light round dots denote samples of s[m] for the upper plot and e[n] for the lower plot.)

7. Figure Illustration of computation of the autocorrelation function for a finitelength sequence. (Square dots denote samples of s[n + m], and light round dots denote samples of s[n].)

8. Figure Illustration (for p = 5) of computation of prediction error for the covariance method. (In upper plot, square dots denote samples of hA[n −m], and light round dots denote samples of s[m].)

9. Figure Illustration of computation of covariance function for a finite-length sequence. (Square dots denote samples of s[n − k] and light round dots denote samples of s[n − i].)

10. Figure Normalized mean-squared prediction error V(p) as a function of model order p in Example 11.2.

11. Figure 11. 11 (a) Windowed voiced speech waveform
Figure (a) Windowed voiced speech waveform. (b) Corresponding autocorrelation function (samples connected by straight lines).

12. Figure (a) Comparison of DTFT and all-pole model spectra for voiced speech segment in Figure 11.11(a). (b) Normalized prediction error as a function of p.

13. Figure 11. 13 (a) Windowed unvoiced speech waveform
Figure (a) Windowed unvoiced speech waveform. (b) Corresponding autocorrelation function (samples connected by straight lines).

14. Figure (a) Comparison of DTFT and all-pole model spectra for unvoiced speech segment in Figure 11.13(a). (b) Normalized prediction error as a function of p.

15. Figure Zeros of prediction error filters (poles of model systems) used to obtain the spectrum estimates in Figure

16. Figure 11.16 Spectrum estimation for a sinusoidal signal.

17. Figure 11.17 Equations defining the Levinson–Durbin algorithm.

18. Figure Comparison of the Levinson–Durbin algorithm and the algorithm for converting from k-parameters of a lattice structure to the FIR impulse response coefficients in Eq. (11.85).

19. Figure 11.19 Signal flow graph of prediction error computation.

20. Figure Signal flow graph of lattice network implementation of pth-order prediction error computation.

21. Figure 11.21 All-pole lattice system.

22. Figure P11.9

23. Figure P

24. Figure P

25. Figure P

26. Figure P11.17

27. Figure P11.18

28. Figure P11.19

29. Table 11.1 PREDICTION COEFFICIENTS FOR A SET OF LINEAR PREDICTORS

30. Figure P

31. Figure P11.25-2 Lattice structure for 2nd-order system

32. Figure P11.25-3 Lattice structure for 4th-order system

33. Figure P

34. Figure P