EC4410_FY04

EC 4410 Speech Signal Processing

Text and References:

Text: Speech and Audio Signal Processing, Gold & Morgan, 1999, Wiley and Sons

Ref 1: Discrete Time Processing of Speech Signals, J. Deller et al, Macmillan, 1993.

Ref 2: Digital Processing of Speech Signals, Rabiner & Schafer, PrenticeHall, 1978.

Ref 3: Speech Communication Human and Machine, O'Shaughnessy, AddisonWesley, 1987.

Course Objectives:

To introduce students to the treatment of digital modeling, analysis, and synthesis of speech signals, speech and speaker recognition.

Examinations, Homework, and Grading:

1 inclass test, worth: 25% Speech Technology Brief/report: 25%

Speech related projects: 40% Class participation: 10%; no final examination

Computer Usage:

Students will be required to use PCs for projects assigned during the course. Some of the lectures may be held in the computer lab.

Students will be required to have a microphone available to digitize their speech data.

Preliminary Course Outline:

!Introduction to speech signal processing and its applications. (Chapters 1-4 and research papers)

!Review of fundamentals of DSP; transform representation of signals and systems, digital filters (FIR, IIR). (Chapters 6 & 7, Ref 1)

! Digital models for the speech signal; speech production, speech characteristics, digital models for speech signals. (Chapters 2&3, Ref 1)

!Timedomain models speech processing; shorttime energy, shorttime autocorrelation, speech versus silence discrimination, pitch period estimation. (Chapter 30)

!Frequency domain speech processing; shorttime Fourier analysis, pitch detection, analysis & synthesis. (Ref 1)

!Cepstral techniques. (Chapter 20)

! Linear predictive coding; LPC equations, relations between speech parameters, applications of LPC parameters. (Chapter 21)

!Pattern classification (Chapters 8&9)

!Topics in Coding (vocoder, MPEG)

!Selection of recent developments in speech recognition (Text, chapters 22-28, ref.)

Projects:

Several (3 to 4) computerbased projects will be assigned during the quarter to illustrate and extend the concepts covered in the classroom. You are encouraged to discuss the projects with other students. However, data collection and code implementation are to be done on an individual basis, and written reports should contain individual work only. Data and code from other students in the class are not to be used in the reports. Code found on the web and used for the projects is to be explicitly identified and properly referenced in the report.

Speech Technology Brief/Report:

A topic related to specific applications of speech processing will be assigned to each student enrolled in the course at the beginning of the quarter. Each student will:

1)Conduct a thorough search of the specific area (including database searches) to gather background and up-to-date information on the topic.

2)Investigate specific domains of applications, advantages, drawbacks of speech technology in that application,

3)Present findings in a ~30mn brief to the rest of the class, the brief will be to be supported by information presented on powerpoint or overhead transparencies.

4)Turn in a 10 pages max written report summarizing finding, with supporting references included at the back of the report.

Grading will be based on the following:

  1. Thorough research of the area assigned,
  1. Completeness and accuracy of the information presented,
  2. Professional presentation of the issues brought up in the articles assigned,
  3. Quality of the report,
  4. feedback from other students in the class (50%)

Class participation grade:

Will be based on class participation during the quarter.