CSVTU BE VII Semester ET&T Speech Signal Processing Syllabus

Chhattisgarh Swami Vivekanand Technical University, Bhilai

Semester : VII Branch: Electronics & Telecommunication

Subject: Speech Signal Processing

Total Theory Periods: 40 Total Tutorial Periods: 12

Total Marks in End Semester Examination: 80

Minimum number of Class tests to be conducted: Two

UNIT – I

Speech: Production, Perception and Acoustic-Phonetic Characterization:

 

Introduction, Speechproduction

process, Time and frequency domain representation of speech, Speech sounds and features, The

 

vowels, Diphthongs, Semivowels, Nasal Consonants, Unvoiced Fricatives, Voiced Fricatives, Voiced &

 

Unvoiced Stops, Acoustic-Phonetic Approach to Speech Recognition, Statistical Pattern-Recognition Approach

 

to Speech Recognition, AI Approaches to Speech Recognition, Neural Networks and their Application to

 

Speech Recognition.

 

UNIT – II

Spectral Analysis of Speech:

 

Short time Fourier analysis, filter bank design, speech coding, subband coding

of speech, transform coding, channel vocoder, formant vocoder, cepstral vocoder, vector quantizer coder.

 

UNIT – III

Speech Synthesis:

 

Pitch extraction algorithms, Gold Rabiner pitch trackers, autocorrelation pitch trackers,

voice/unvoiced detection, homomorphic speech processing, homomorphic systems for convolution, complex

 

cepstrums, pitch extraction using homomorphic speech processing

 

UNIT – IV

Automatic speech recognition systems:

 

Isolated word recognition, connected word recognition, large

vocabulary word recognition systems, pattern classification, DTW, HMM, speaker recognition systems,

 

speaker verification systems, speaker identification systems.

 

UNIT – V

Hidden Markov Models:

 

Discrete-Time Markov Processes, Extensions to HMMs, Coin-toss Models, The Urnand-

Ball Model, Elements of an HMM, HMM generator of observations. Three Basic problems for HMMs and

 

their solutions, Probability Evaluation, ‘Optimal’ State sequence, Parameter estimation, Re-estimation

 

procedure. HMM types, continuous observation densities in HMMs, Autoregressive HMMs, Variants on HMM

 

structures, Inclusion of Explicit State Duration Density in HMMs, Optimization Criterion – ML, MMI and MDI,

 

Comparisons of HMMs.

 

Name of Text Books:

1. Fundamentals of Speech Recognition, Rabiner L. and Juang B., Pearson Education

2.

 

Owens F.J., ‘‘Signal Processing of Speech’’, Macmillan New Electronics

 

Names of Reference Books:

1. Speech and Language Processing, Jurafsky, Pearson Education

2. Discrete Time Speech Signal processing: Principles and Practice, Quatieri, Pearson Education

3. Saito S. & Nakata K., “Fundamentals of Speech Signal Processing”, Academic Press

4. Thomas Parsons, “Voice and Speech Processing”, McGraw Hill Series

Leave a Comment