Unit 1 Principle characteristics of speech

9 lecture hours

• linguistic information
• speech and hearing
• speech production mechanism
• articulatory phonetics: manner of articulation and place of articulation
• acoustic phonetics: spectrograms
• wide-band and narrow-band spectrograms
• acoustic characteristic of speech
• statistical characteristics of speech
• speech production models

Unit 2 Speech analysis and synthesis systems

9 lecture hours

• quantization and coding
• spectral analysis
• spectral structure of speech
• short-time autocorrelation
• average magnitude difference function
• short time Fourier transform
• window function
• sound spectrogram
• homomorphic processing
• properties of complex cepstrum
• complex cepstrum of voiced and unvoiced speech
• mel-frequency cepstral coefficients
• filter bank and zero crossing analysis
• analysis –by-synthesis
• pitch extraction

Unit 3 Linear predictive analysis

9 lecture hours

• Basic principles of linear predictive analysis
• autocorrelation method and covariance method
• computation of gain for the model
• prediction error signal
• frequency domain interpretation of LP analysis
• frequency domain interpretation of mean-squared prediction error
• applications of LPC parameters

Unit 4 Speech coding

• reversible coding
• irreversible coding and information rate distortion theory
• coding in time domain: PCM, ADPCM
• adaptive predictive coding
• coding in frequency domain: sub band coding
• adaptive transform coding
• vector quantization
• code excited linear predictive coding (CELP). 09 Subject Code: 3724119 w.e.f