Princeton University Computer Science Department

Computer Science 597E AdTopCS: Digital Speech Processing M. Rahim

Fall 2004

Course Summary

Human/machine communication is one of the most fascinating areas of Artificial Intelligence. Inspired by scientists and the movie industry, communicating with machines by voice is beginning to be part of every day life. This course provides a graduate-level study of how computers talk, recognize and understand speech. Topics include speech production, perception, acoustic and language modeling, signal processing for feature extraction, hidden Markov modeling of speech, connected speech recognition, and spoken language dialog. Students will have the opportunity to build software agents that can speak and recognize spoken commands. Grades for this course will be split equally between homework project assignements (50%) and final course project (50%).

Lectures

Lecture 1: Introduction into Speech Processing and Human/Machine Communication

Lecture 2: Signal Processing Methods for Speech Analysis

Lecture 3: Linear Predictive Coding and Analysis of Speech

Lecture 4: Vector Quantization and Dynamic Search

Lecture 5: Vector Quantization and Hidden Markov Models

Lecture 6: Hidden Markov Models for Speech Processing

Lecture 7: Language Modeling and Finite State Automata

Lecture 8: Speech Processing Technologies and Applications

Lecture 9: Natural Language Processing and Parsing

Lecture 10: Machine Learning for Spoken Language Processing

Lecture 11: Spoken and Multimodal Human/Machine Dialog

---

Administrative Information

Lectures: Fridays 1.30pm-4.20pm, Room: 302

Professor: Mazin Rahim - CS Building - 258-; mazin@cs.princeton.edu

Graduate Coordinator: Melissa Lawson - 310 CS Building - 258-5387 mml@cs.princeton.edu

Teaching Assistants: TBA