No. of Pages: 2

APJ ABDUL KALAM TECHNOLOGICAL UNIVERSITY
FIRST SEMESTER M.TECH DEGREE EXAMINATION, DECEMBER 2017

Electronics & Communication Engineering
1. Signal ProcRSing
2. VLSI and Embedded Systems
01೯೮6311 Speech Signal Processing

Answer any twofull questions from each part
Limit answers to the required points.
Max. Marks 60 Duration: 3 hours
PARTA

1.3. Analyze the acoustic modeling of the vocal tract in speech processing. (5)

b. Describe the state ofthe Glottis during the pronunciation ofthe following phonemes:
i) W, (ii) /b/, (iii) /2/ (iv) 15/ (2) c. Determine Fo ofthe following signal ifthe signal

is sampled at 22050 Hz. (2)
പം
a. Explain frequency domain analysis of speech signals using filter banks. (2)

b. Explain how cepstral analysis is used in speech signal processing, with relevant block
diagram. (5)

c. Suppose uniform filter bank analysis is used to extract the parameters of a speech
segment. If the bandwidth of each filter is 100 Hz and speech signal is recorded with
sampling frequency 12 KHz, determine the required number offilters to cover the entire

spectrum of the speech segment. (2)

3. . Describe the use of zero crossing rate and autocorrelation function determination in time-
domain analysis of speech signals. (5)

b. Differentiate between narrowband and wideband spectrograms. (4)

PART B