Speech Processing

SIPCom8-4

Speech Processing

The course is given by Per Rubak (PR), Mads Græsbøll Christensen (MGC), Søren Holdt Jensen (SHJ), Chunjian Li (CL), and Zheng-Hua Tan (ZT). The course coordinator is Zheng-Hua Tan.

Literature:

Deller, Hansen, Proakis, Discrete-Time Processing of Speech Signals, 2nd. edition, Wiley-IEEE Press, 1999.

---------------------------------------------

Lecture 1:

Time and place:

Tuesday, February 7th 2006

Lecture, 12:30 to 14:00 in A4-106

Problem solving, 14:14 to 16:15

Topics: Fundamentals of speech science, model of speech production

Lecturer: ZT

Preparation:

Slides for MM1

Literature: Deller, Hansen, Proakis, “Discrete-Time Processing of Speech Signals”, chapter 2 and 3, (pp. 81-85, 99-146, 151-201)

Exercise:

Speech files associated with the textbook and Speech files for this exercise. The tool - Speech Filing System.

Exercise for MM1.

---------------------------------------------

Lecture 2:

Time and place:

Tuesday, February 14th 2006

Lecture, 12:30 to 14:00 in A4-106

Problem solving, 14:14 to 16:15

Topic:

Speech perception

Lecturer: PR

Preparation:

Slides

Literature:

Exercise:
Solution

---------------------------------------------

Lecture 3:

Time and place:

Tuesday, February 21st 2006

Lecture, 12:30 to 14:00 in A4-106

Problem solving, 14:14 to 16:15

Topics:

Fundamentals of linear prediction-based speech coding.

Lecturer:
MGC

Literature:
W. B. Kleijn and K. K. Paliwal, Eds., "Speech Coding and Synthesis", 1995, Chapter 1 (pp. 3-47) and 3 (pp. 79-119).
Excercises:

Click here.

---------------------------------------------

Lecture 4:

Time and place:

Thursday, February 28th 2006

Lecture, 12:30 to 14:00 in A4-106

Problem solving, 14:14 to 16:15

Topics:

Parametric speech and audio coding, relation to vector quantization, efficient implementation of perceptual distortion measures.

Lecturer:

MGC

Literature:
M. G. Christensen and S. H. Jensen, "On perceptual distortion minimization and nonlinear least-squares frequency estimation", IEEE Trans. on Audio, Speech and Language Processing, Volume 14, Issue 1, Jan. 2006, pp. 99-109 (pdf).

Excercises:

Click here.

---------------------------------------------

Lecture 5:

Time and place:

Tuesday, March 7^th 2005

Lecture, 12:30 to 14:00 in A4-106

Problem solving, 14:14 to 16:15

Topics:
Rate-distortion optimization, rate-distortion optimized LPC.

Lecturer:

MGC

Literature:

P. Prandoni and M. Vetterli, "R/D optimal linear prediction", IEEE Trans. on Speech and Audio Processing, Volume 8, Issue 6, Nov. 2000, pp. 646-655 (pdf).

Excercises:

Click here.

---------------------------------------------

Lecture 6:

Time and place:

Tuesday, March 21st 2006

Lecture, 12:30 to 14:00 in NJ14 3-119

Problem solving, 14:14 to 16:15

Topics:

Speech enhancement, Single-Channel Spectral Subtraction based methods

Lecturer: SHJ / CL

Preparation:

Slides for MM6
Literature:

Deller, Hansen, Proakis, “Discrete-Time Processing of Speech Signals”, chapter 8

Supplementary Literature:

D. Ealey, H. Kelleher, and D. Pearce, "Harmonic Tunneling: tracking non-stationary noise during speech", Eurospeech 2001

R. Martin, "Noise Power Spectral Density Estimation Based on Optimal Smoothing and Minimum Statistics", Trans. Speech and Audio Processing, vol. 9, no. 5, July 2001, pp. 504-514

Exercise:

Given at the lecture

---------------------------------------------

Lecture 7:

Time and place:

Tuesday, March 28th 2006

Lecture, 12:30 to 14:00 in NJ14 3-119

Problem solving, 14:14 to 16:15

Topic:

Speech enhancement, Multi-Channel methods

Lecturer: SHJ / CL

Preparation:

Slides for MM7

Literature:

Deller, Hansen, Proakis, “Discrete-Time Processing of Speech Signals”, chapter 8

Exercise:

Given at the lecture

---------------------------------------------

Lecture 8:

Time and place:

Tuesday, April 4th 2006

Lecture, 12:30 to 14:00 in A4-108 (Note the room!)

Problem solving, 14:14 to 16:15

Topic:

Speech synthesis & Speech Recognition (DTW)

Lecturer: ZT

Preparation:

Slides for MM8

Literature:

D. O'Shaughnessy, "Interacting with Computers by Voice: Automatic Speech Recognition and Synthesis", Proceedings of the IEEE, 91 (9), Sept. 2003. (PDF. Password for opening the PDF file is the name of your specialisation excluding semester number, all in lowercase)
Deller, Hansen, Proakis, “Discrete-Time Processing of Speech Signals”, chapter 11. Alternative: Sakoe, H. and Chiba, S., "Dynamic programming algorithm optimization for spoken word recognition", IEEE Trans. Acoustics, Speech, and Signal Processing, 26(1), 1978. (PDF)

Exercise:

Exercise 1: Online demo.

Exercise 2: Use Matlab to implement Dynamic Time Warping to compare speech signals

---------------------------------------------

Lecture 9:

Time and place:

Tuesday, April 11th 2006

Lecture, 12:30 to 14:00 in NJ14 3-119 (Note the room!)

Problem solving, 14:14 to 16:15

Topic:

Speech recognition, HMM

Lecturer: ZT

Preparation:

Slides for MM9

Literature:

Deller, Hansen, Proakis, “Discrete-Time Processing of Speech Signals”, chapter 12. Alternative: Rabiner, L.R., "A tutorial on hidden Markov models and selected applications in speech recognition", Proceedings of the IEEE, 77 (2), 1989, pp. 257 - 286. (PDF. Password for opening the PDF file is the name of your specialisation excluding semester number, all in lowercase)
Steve Young, et al., "The HTK Book" (PDF)

Exercise:

Exercise 1: Hidden Markov model and Viterbi decoding.

Exercise 2: HTK demo.

---------------------------------------------

Lecture 10:

Time and place:

Tuesday, April 18th 2006

Lecture, 12:30 - 14.00 in A4-106 (Note the room!)

Problem solving, 14.14 to 16:15

Topic:

Large Vocabulary Continuous Speech Recognition (LVCSR) and More

Lecturer: ZT

Preparation:

Slides for MM10

Literature:

Steve Young, "A review of large-vocabulary continuous-speech", IEEE Signal Processing Magazine, Sep 1996 (PDF. Password for opening the PDF file is the name of your specialisation excluding semester number, all in lowercase). Alternatively, Huang, Acero and Hon, Spoken Language Processing, Chapter 9.

Exercise:

Given at the lecture.