�@

�@

�@

�@

�@

�@

�@

�@

�@

�@

�@

�@

�@

�@

�@

�@

Administrative Information - Fall 1999

• Instructor: Alen Docef ([email protected]) Office hours: two hours before class, CICSR 361 Phone: 822-4307 Fax: 822-5949
• Class hours: Monday, 15:30-18:30, MCLD 208

1. DSP Fundamentals (1 week)
   • Linear shift invariant systems
   • Fourier and Z transforms
   • Digital filters
   • Sampling and reconstruction
      
2. Speech Models (1 week)
   • Acoustic theory of speech production
   • Tube models for speech generation
   • Properties of speech waveforms
   • Digital models and basic problems of speech processing
      
3. Time-Domain Analysis Methods (1 week)
   • Peak, energy and zero-crossing measurements
   • Pitch period estimation
   • Auto-correlation analysis
      
4. Short-Time Spectrum Analysis Methods (1 week)
   • Definitions, filterbanks, computation, sound spectrograms
   • Window effects
   • Pitch detection and vocoders
      
5. Homomorphic Speech Processing (1 week)
   • Cepstrum of speech signals
   • Pitch detection
   • Formant analysis and applications
      
6. Linear Prediction Analysis Methods (1 week)
   • Formant analysis
   • Spectrum analysis and determination
   • Analysis-synthesis systems
      
7. Speech Recognition (2 weeks)
   • Time-warping methods
   • Hidden Markov models
   • Neural networks
      
8. Text-To-Speech Synthesis (1 week)
   • LPC synthesizer
   • Limited vocabulary encoding
   • Text-to-allphone rules
      
9. Digital Audio (3 weeks)
   • What is digital audio?
   • Audio perceptual modeling
   • Audio recording and transmission
   • Audio broadcasting
      

Text: Speech and Audio Signal Processing, by B. Gold and N. Morgan.

Grading:

• Homework: 15%
• Midterm Exam #1: 30%
• Midterm Exam #2: 30%
• Project: 25% (10% presentation, 10% demo, 5% written report).

Miscellaneous:

• lms.m
• Oversampling assignment fun