4. a. 061۷۵ the expression for autocorrelation matrix in linear prediction analysis of speech
signals. (5)
b. State the property of the Toeplitz matrix that makes it suitable for using recursive
algorithms [ike the Levinson Durbin algorithm.

c. Write the Levinson Durbin equations used in solving autocorrelation matrix. (3)
5. a. Explain the application of Hidden Markov model in speech analysis. (6)
b. Describe the sinusoidal model of speech. (3)
6. a. Derive the Covariance matrix of LPC analysis. (5)
b. Describe Gaussian Mixture Modeling of speech. (4)
PART C
a. Explain how sub-band coding is performed in speech coding. (6)
b. Describe any one vocoder based on the CELP algorithm. (6)
8. a. With the help of a block diagram, explain how a Text-to-speech system is implemented.
(6)
b. Describe how segmentation is achieved in speech recognition systems. (4)

௦. An audio signal is recorded using sampling frequency FS = 8 kHz, encoded with 16 bit and
recorded in mono. How much memory is required to store 100ms signal in PCM WAV
format? (2)

9. a. Distinguish between adaptive transform coders and harmonic coders. (6)

b. Explain the issues faced in language identification and voice transmission over the
intemet. (6)