SAMPLE COURSE OUTLINE CKCS 903 FUNDAMENTALS OF SPEECH RECOGNITION

Transcription

SAMPLE COURSE OUTLINE CKCS 903 FUNDAMENTALS OF SPEECH RECOGNITION
SAMPLE COURSE OUTLINE
CKCS 903
FUNDAMENTALS OF SPEECH RECOGNITION
This is a sample course outline only. It should not be used to plan assignments or purchase textbooks.
A current version of the course outline will be provided by the instructor once the course begins.
Every effort will be made to manage the course as stated. However, adjustments may be necessary at the
discretion of the instructor. If so, students will be advised and alterations discussed in the class prior to
implementation.
It is the responsibility of students to ensure that they understand the University’s policies and procedures,
in particular those relating to course management and academic integrity
COURSE DESCRIPTION
This course covers the fundamentals of speech recognition: signal processing and analysis methods for
speech recognition, different pattern comparison techniques, speech recognition system design and
implementation issues, basic principles of Hidden Markov Model, connected work model, applications of
automatic speech recognition (ASR) such for civil and military applications. Students will learn the
current state-of-art of speech recognition: digital microphone array for distant speech recognition,
hardware for real-time speech recognition using a Liquid State Machine, Computer-Aided Digital Note
Taking System on Physical Book, Mathifier - Speech recognition of math equations, speech recognition
using fuzzy logic, Human recognition through RFID A distinct application of speech processing and so
on.
COURSE OBJECTIVE/LEARNING OUTCOMES
The main objectives of this course are:




To provide students with opportunities to develop the fundamentals in speech recognition.
To identify and teach if there is any specific topic or application the students of different
discipline want to learn and adjust the course outline accordingly.
To provide students with opportunities to learn programming in MATLAB for digital speech
recognition.
To discuss the current research trends in speech recognition.
Sample Course Outline
Fundamentals of Speech Recognition
Fall 2012
CKCS903
Page 1 of 4
TEXTBOOK AND READING LISTS
This is a sample course outline only. It should not be used to purchase textbooks. A current version
of the course outline will be provided by the instructor once the course begins.
Readings and Related Material:



Nejat Ince, Digital Speech Processing: Speech Coding, Synthesis and Recognition, Kluwer
Academic Publishers
L.R. Rabiner and B-H. Juang, Fundamentals of Speech Recognition, Prentice-Hall Signal
Processing Series
Vinay K. Ingle and John G. Proakis, Digital signal processing using MATLAB.
Research articles of different journals and conferences to learn the current state-of-art in digital speech
recognition
COURSE STRUCTURE AND ORGANIZATION:
Each class will consist of two components: A lecture that covers theory and an overview of practical
applications of concepts; a lab session with MATLAB functionalities and other programming languages
for speech recognition.
SCHEDULE OF TOPICS:
Week
Topic
Details
WK 1
Introduction
Introduction to Speech Recognition, a brief history and
applications of speech recognition, approaches to
automatic speech recognition by machines
WK 2
Analysis Method for
Speech Recognition
Bank-of-filters front-end processor, linear predictive
coding model, vector quantization, auditory-based spectral
analysis models
WK 3
Pattern Comparison
Technique
Speech detection, mathematical and perceptual
considerations of distortion measures, spectral-distortion
measures, time alignment and normalization
MATLAB, Industry
available software
Discuss MATLAB functionalities and demonstrate
available software for speech recognition
WK 4
Assignment/problem set posted
WK 5
Speech Recognition
System Design
WK 6
Speech Recognition
Models
Wk 7
Review
Sample Course Outline
Fundamentals of Speech Recognition
Source coding techniques to recognition, template training
method, discriminative method, speech recognition in
adverse environment and so on.
Hidden Markov Model, Connected Word Model,
Continuous large vocabulary
Assignment/problem set submission
Fall 2012
CKCS903
Page 2 of 4
EVALUATION:
This is a non-credit course. However, students are required to submit an assignment or problem set for
evaluation.
MISSED TERM WORK OR EXAMINATIONS
Students are expected to complete all assignments, tests, and exams within the time frames and by the
dates indicated in this outline. Exemption or deferral of an assignment, term test, or final examination is
only permitted for a medical or personal emergency or due to religious observance. The instructor must
be notified by e-mail prior to the due date or test/exam date, and the appropriate documentation must be
submitted. For absence on medical grounds, an official student medical certificate, downloaded from the
Ryerson website at http://www.ryerson.ca/senate/forms/medical.pdf or picked up from The Chang
School at Heaslip House, 297 Victoria St., Main Floor, must be provided. For absence due to religious
observance, visit http://www.ryerson.ca/senate/forms/relobservforminstr.pdf to obtain and submit the
required form.
PLAGIARISM
The Ryerson Student Code of Academic Conduct defines plagiarism and the sanctions against students
who plagiarize. All Chang School students are strongly encouraged to go to the academic integrity
website at www.ryerson.ca/academicintegrity and complete the tutorial on plagiarism.
ACADEMIC INTEGRITY
Ryerson University and The Chang School are committed to the principles of academic integrity as
outlined in the Student code of Academic conduct. Students are strongly encouraged to review the student
guide to academic integrity, including penalties for misconduct, on the academic integrity website at
www.ryerson.ca/academic integrity and the Student code of Academic conduct at
www.ryerson.ca/senate/policies.
RYERSON STUDENT EMAIL
All students in full and part-time graduate and undergraduate degree programs and all continuing
education students are required to activate and maintain their Ryerson online identity at
www.ryerson.ca/accounts in order to regularly access Ryerson’s E-mail (Rmail), RAMSS, my.ryerson.ca
portal and learning system, and other systems by which they will receive official University
communications.
COURSE REPEATS:
Senate GPA policy prevents students from taking a course more than three times. For complete GPA
policy see policy no. 46 at www.ryerson.ca/senate/policies.
Sample Course Outline
Fundamentals of Speech Recognition
Fall 2012
CKCS903
Page 3 of 4
RYERSON ACADEMIC POLICIES
For more information on Ryerson’s academic policies, visit the Senate website at www.ryerson.ca/senate.
Course Management Policy No. 145
Student Code of Academic conduct No. 60
Student code of non-Academic Conduct No. 61
Examination Policy No. 135
Policy on Grading, Promotion, and Academic Standing Policy No. 46
Undergraduate Academic consideration and Appeals Policy No. 134
Accommodation of Student Religious Observance Obligations Policy no. 150
Sample Course Outline
Fundamentals of Speech Recognition
Fall 2012
CKCS903
Page 4 of 4

Similar documents