Introducing the INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING - - PDF document

▶
introducing the institute for signal and information
SMART_READER_LITE
LIVE PREVIEW

Introducing the INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING - - PDF document

Sep 17, 2023 153 likes •336 views

INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING Introducing the INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING located at Mississippi State University Department of Electrical and Computer Engineering Box 9571, Mississippi State, Mississippi

slide-1
SLIDE 1

INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING

I S I P I S I P

s p ee c h s p ee c h

Introducing the INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING located at Mississippi State University Department of Electrical and Computer Engineering Box 9571, Mississippi State, Mississippi 39762 Tel: 601-325-3149 Fax: 601-325-3149 email: picone@isip.msstate.edu MISSION STATEMENT Mississippi State University for over 100 years has had a mission of being a center of excellence in the State of Mississippi for:

  • Learning — to enhance the intellectual development of its students
  • Research — to extend the present limits of knowledge
  • Service — to apply its research to improve the lives of people

The Institute for Signal and Information Processing (ISIP) offers a multidisciplinary program focused

  • n

the development

  • f

next generation information processing techniques. Research at ISIP is centered on intelligent information processing, perhaps the most important technology of the next century. ISIP draws upon a wide range of research experience in areas such as signal processing, communications, natural language, database query, intelligent systems, and discrete controls. Its present vision is to develop systems capable

  • f

intelligent interactions with users by the integration of a multiplicity of interface technologies including speech, natural language, database query, and imaging.

slide-2
SLIDE 2

INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING OCTOBER 2, 1997 TELECOMMUNICATIONS PAGE 1 OF 16

I S I P I S I P

s p ee c h s p ee c h

Anthony Skjellum High Performance Computing Computer Science / ERC

Joe Picone Signal Processing

  • Inst. for Signal and Info. Proc.

Robert J. Moorhead Image Processing ERC SIGNAL PROCESSING RESEARCH AT MISSISSIPPI STATE UNIVERSITY IS MULTIDISCIPLINARY Stephen E. Saddow Semiconductor Technology EMRL Victor A. Rudis Forestry Imaging USFS

Communications Laboratory

  • Elect. and Comp. Eng.

Bud Rizer Assistive Technologies T.K. Martin Center

slide-3
SLIDE 3

INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING OCTOBER 2, 1997 TELECOMMUNICATIONS PAGE 2 OF 16

I S I P I S I P

s p ee c h s p ee c h

isip00 (fileserver, router, and domain server):

  • Sun SPARC 5
  • 70 MHz MicroSPARC II
  • 32 Mbytes RAM, 1 Gbyte local disk
  • 2 ethernets (for routing)
  • 60 Gbytes magnetic disk (Seagate Elite)

Exabyte 10h Tape Library

  • 8 mm tapes
  • 70 Gbyte capacity
  • 140 Gbytes compressed

Outside World (hub #0):

  • Allied Telesyn MR 820T
  • 10BaseT 8 port hub (10 Mbits/sec)
  • Cat-5 Unshielded Twisted Pair
  • 155 Mbits/sec ATM (campus)

isip01 (compute server):

  • Sun SPARC 20-512
  • Two 50 MHz SuperSPARC Processors
  • 192 Mbytes RAM, 1 Gbyte local disk

isip02 (demo machine):

  • Sun Sparc 5
  • 70 MHz MicroSPARC II
  • 32 Mbytes RAM, 1 Gbyte local disk
  • T1 Telecom Interface

datlink 0 and datlink 1 (audio):

  • Townshend DAT-Link+
  • 16-bit digital audio
  • AES/EBU and SP-DIF

Sharp JX-325 Color Scanner:

  • one-pass 24-bit color scan
  • 300 dpi native mode

Domain: isip.msstate.edu

isip03 and isip05 (compute server):

  • dual Pentium Pro
  • 200 MHz Processor
  • 256 Mbytes RAM, 1Gbyte local disk

isip04 and isip06 (laptops):

  • Samsung Sens 810, Toshiba Tecra 500 CDT
  • 133 MHz Pentium Processor
  • 40 Mbytes RAM, 2 Gbyte local disk

ncd20c00 (clients):

  • NCD Xterms
  • 16-bit audio
slide-4
SLIDE 4

INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING OCTOBER 2, 1997 TELECOMMUNICATIONS PAGE 3 OF 16

I S I P I S I P

s p ee c h s p ee c h

ISIP’S FOCAL PROJECT

  • An Integrated Services Transactions Processor That Supports

Advanced Telecommunications Interfaces such as an Asynchronous Transfer Mode (ATM) Digital Communications Link

Example: Telephone-Based Natural Language Query of Entertainment Archives Customer: “Give me all movies, uh, make that only the recent movies, directed by Martin Scorsese and starring Robert DeNiro, and oh, by the way, make that movies about gangsters only.” Computer: We have three titles available (the titles of the movies are shown on the television screen with real-time video of promo clips from each movie below the title). Please select a movie. Customer: “That one with the three guys looks good, I’ll take that one. I want it to start at 8:00 PM tomorrow.” Computer: (The promo clip for the selected movie starts playing on the television.) The movie titled GoodFellas starring Robert DeNiro and directed by Martin Scorsese will be delivered for viewing on your television on Thursday, September 25 starting at 8:00 PM. Thank you for using ISIP’s Entertainment

  • Server. Good-bye.

Local Central Office ATM (160 Mbps)

  • Voice
  • Video
  • Data (X Windows)

Unix Multiprocessor (Sparcstation 2000):

  • 8 Processors
  • 512 Mbytes of memory
  • videotape jukebox
slide-5
SLIDE 5

INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING OCTOBER 2, 1997 TELECOMMUNICATIONS PAGE 4 OF 16

I S I P I S I P

s p ee c h s p ee c h

A T1-BASED DATA COLLECTION SYSTEM FOR SUN/UNIX WORKSTATIONS

slide-6
SLIDE 6

INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING OCTOBER 2, 1997 TELECOMMUNICATIONS PAGE 5 OF 16

I S I P I S I P

s p ee c h s p ee c h Semi-Parser Language Model Tagged Text Natural Language Processing Request Generator Knowledge Extractor Filled Templates Netscape Requests Netscape Knowledge Extraction Flat Parsed Structures Speech Recognition Language Model Text Natural Language Understanding “Show me all the reports from the White House on Healthcare.”

slide-7
SLIDE 7

INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING OCTOBER 2, 1997 TELECOMMUNICATIONS PAGE 6 OF 16

I S I P I S I P

s p ee c h s p ee c h INT_4 FLOAT_8 ANALYZE COMPUTE_STATS COMPUTE CLASS UTIL CLASS UTIL ANALYZER COMPUTER TRANSFORM

  • Detailed performance analysis in a common framework

(Table entries are computation times in usec) Algorithm FFT ORDER 16 64 256 1024 4096 16384 RAD2 20 60 280 1960 10900 97100 RAD4 20 60 250 1800 9720 58220 SRFFT 20 40 160 1060 6140 38100 FHT 20 40 140 640 3800 38100 QFT 20 40 160 880 6560 44020 DITF 20 60 360 2500 12320 104080 PARALLEL IMPLEMENTATIONS OF FAST FOURIER TRANSFORMS

  • Object-oriented software implemented in C++
slide-8
SLIDE 8

INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING OCTOBER 2, 1997 TELECOMMUNICATIONS PAGE 7 OF 16

I S I P I S I P

s p ee c h s p ee c h

BASIC TECHNOLOGY: A PATTERN RECOGNITION PARADIGM BASED ON HIDDEN MARKOV MODELS

Search Algorithms:P Wt i Ot

( ) P Ot Wt

i

( )P Wt

i

( ) P Ot ( )

  • =

Pattern Matching: Wt i P Ot Ot 1 –

… Wt

i

, , ( ) , [ ]

Signal Model: P Ot

Wt

1 –

Wt Wt

1 +

, , ( ) ( )

Recognized Symbols: P S O

( ) max arg

T

P Wt

i

Ot Ot

1 –

… , , ( ) ( )

i

∏

=

Language Model: P Wt i

( )

Prediction

slide-9
SLIDE 9

INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING OCTOBER 2, 1997 TELECOMMUNICATIONS PAGE 8 OF 16

I S I P I S I P

s p ee c h s p ee c h

THE JEIDA JAPANESE COMMON SPEECH DATA CORPUS

. Number of speakers 150 speakers 75 male speakers 75 female speakers Number of items per speaker monosyllables 178 isolated words 35 4-digit sequences 323 items Number of repetitions per item 4 repetitions of each item Range of speaker age 20 yrs. to 60 yrs. Amount of data 120 hours Number of Digital Audio Tapes 76 (120-minute tapes) Total number of utterances 193,800 utterances Number of channels/mic. type 2 (dynamic and condenser mics.) Anticipated size of final corpus (16-bit 16 kHz samples @ 1.0 secs per utterance) 6.5 Gbytes (13 CD-ROMs uncompressed)

slide-10
SLIDE 10

INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING OCTOBER 2, 1997 TELECOMMUNICATIONS PAGE 9 OF 16

I S I P I S I P

s p ee c h s p ee c h

AUTOMATIC GENERATION OF N-BEST PROPER NOUN PRONUNCIATIONS NEURAL NETWORK SOLUTION

  • • •
  • • • • • • • • • • • • • • •
  • • • • • • • • • • • •

10100000001000011101101001 100100111001000 E P S T E I N E P S T AI _ N

CONTEXT LETTER WINDOW INPUT REPRESENTATION INPUT LAYER HIDDEN LAYERS OUTPUT LAYER OUTPUT REPRESENTATION OUTPUT PHONEME

slide-11
SLIDE 11

INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING OCTOBER 2, 1997 TELECOMMUNICATIONS PAGE 10 OF 16

I S I P I S I P

s p ee c h s p ee c h

JAVA APPLETS http://isip.msstate.edu/software/java_system_response Other ISIP Java Applets include:

  • Convolution
  • Frequency Response
  • Nyquist Criterion
  • Analog and Digital Filter Design
  • Compilers and Assembly Code
  • Hidden Markov Model Toolkit
  • Speech Recognition Primer
slide-12
SLIDE 12

INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING OCTOBER 2, 1997 TELECOMMUNICATIONS PAGE 11 OF 16

I S I P I S I P

s p ee c h s p ee c h

SYLLABLE-BASED SPEECH RECOGNITION FOR CONVERSATIONAL TELEPHONE SPEECH

slide-13
SLIDE 13

INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING OCTOBER 2, 1997 TELECOMMUNICATIONS PAGE 12 OF 16

I S I P I S I P

s p ee c h s p ee c h

ECHO CANCELLATION FOR SPEECH RECOGNITION

AUTOMATIC SPEECH RECOGNIZER FOR SPEAKER IDENTIFICATION ANNOUNCER OR A CALLER,

s n ( ) ) a n ( )

s n ( ) SPEAKER, ECHO

(CUE FOR SPEAKER ID)

ECHO CANCELLER HYBRID IN NETWORK

slide-14
SLIDE 14

INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING OCTOBER 2, 1997 TELECOMMUNICATIONS PAGE 13 OF 16

I S I P I S I P

s p ee c h s p ee c h

What Differentiates ISIP Research? ❐ Public Domain Software ❐ Extensive Web Archive ❐ Object-Oriented Signal Processing Software ❐ State-of-the-Art Performance Tasks ❐ Close Industrial Ties ❐ Next-Generation Statistical Models Based

  • n Chaotic Systems

Applicable to acoustic and language modeling Addresses a fundamental barrier in speech understanding

slide-15
SLIDE 15

INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING OCTOBER 2, 1997 TELECOMMUNICATIONS PAGE 14 OF 16

I S I P I S I P

s p ee c h s p ee c h

Algorithms Aravind Ganapathiraju (Ph.D. - 1)

Jule Baca (Ph.D. - 4) Neeraj Deshmukh (Ph.D. - 3) Julie Ngan (M.S. - 1)

Institute for Signal and Information Processing (ISIP) Director: Dr. Joseph Picone Software Jonathan Hamaker (M.S. - 1)

Audrey Le (M.S. - 1) Janna Shaffer (U.G. - 4)

Information Technology Richard Duncan (U.G. - 3)

Nirmala Kalidindi (M.S. - 2) Suresh Balakrishnam (M.S. - 1) New Hires (U.G. - 3)

slide-16
SLIDE 16

INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING OCTOBER 2, 1997 TELECOMMUNICATIONS PAGE 15 OF 16

I S I P I S I P

s p ee c h s p ee c h

Joseph Picone Associate Professor Department of Electrical and Computer Engineering

Mississippi State University Phone: (601) 325-3149 Box 9571 Fax: (601) 325-2298 Mississippi State, MS 39762 Email: picone@isip.msstate.edu

Education

Ph.D. in Electrical Engineering, Illinois Institute of Technology, December 1983 M.S. in Electrical Engineering, Illinois Institute of Technology, May 1980 B.S. in Electrical Engineering, Illinois Institute of Technology, May 1979

Areas of Research

Speech Understanding, Digital Signal Processing, and Pattern Recognition.

Experience Summary

  • Dr. Picone primary interests are in the area of new statistical approaches to speech
  • understanding. He has founded a speech research laboratory at Mississippi State

University that conducts research into a number of related areas. (For more informa- tion, please check http://www.isip.msstate.edu). Research support has included projects with Texas Instruments, the Linguistic Data Consortium, ARPA’s Spoken Language Systems program, and DoD.

  • Dr. Picone recently served as Data and Systems Coordinator for the 1997 Summer

Workshop on Large Vocabulary Speech Recognition hosted by the Center for Lan- guage and Speech Processing at Johns Hopkins University. During this workshop, he also served as a senior member of a team dedicated to syllable-based speech pro-

  • cessing. Under his guidance, the workshop was extremely successful as all four

teams participating in the workshop posted statistically significant improvements on the state of the art.

  • Dr. Picone is currently a Senior Member of the IEEE and a Professional Engineer reg-

istered in the State of Texas. He is also an Associate Editor for the IEEE Signal Pro- cessing Magazine and the IEEE Transactions on Speech and Audio Processing, and has served as a reviewer for numerous organizations including NSF. He was previ-

  • usly employed at Texas Instruments as a Senior Member of Technical Staff and at

AT&T Bell Laboratories. He is also a former Adjunct Professor at University of Texas at Dallas and Illinois Institute of Technology. He has previously conducted research in medium and low data rate speech compression. Dr. Picone has published more than 85 papers in the area of speech processing and has been awarded 8 patents.

slide-17
SLIDE 17

INSTITUTE FOR SIGNAL AND INFORMATION PROCESSING OCTOBER 2, 1997 TELECOMMUNICATIONS PAGE 16 OF 16

I S I P I S I P

s p ee c h s p ee c h

Recent Significant Publications

Journal Articles:

  • 1. N. Deshmukh and J. Picone, “Automatic Generation of N-Best Pronunciations of

Proper Nouns,” submitted to the IEEE Transactions on Speech and Audio Pro- cessing, November 1996.

  • 2. J. Picone, T. Staples, K. Kondo and N. Arai, “Kanji to Hiragana Conversion Based
  • n a Length Constrained N-Gram Analysis,” accepted for publication in the IEEE

Transactions on Speech and Audio Processing, Fall 1996.

  • 3. J. Picone, W.J. Ebel, and N. Deshmukh, “Automated Speech Understanding: The

Next Generation,” in Digital Signal Processing Technology, Vol. CR57, pp. 101- 114, 1995. Conferences:

  • 4. N. Deshmukh, J. Ngan, J. Hamaker, and J. Picone, “An Advanced System to Gen-

erate Multiple Pronunciations of Proper Nouns,” Proceedings of the IEEE Interna- tional Conference on Acoustics, Speech, and Signal Processing, Munich, Germany, vol. 2, pp. 1467-1470, April 1997.

  • 5. J.J. Godfrey,
  • A. Ganapathiraju,

and

  • J. Picone,

“Microsegment Modeling for Speech Recognition,” Proceedings of the IEEE International Conference on Acoustics, Speech, and Signal Processing, Munich, Germany,

  • vol. 3,
  • pp. 1755-1758, April 1997.
  • 6. A. Ganapathiraju and J. Picone, “Echo Cancellation For Evaluating Speaker Iden-

tification Technology,” Proceedings of IEEE Southeastcon, pp. 100-102, Blacks- burg, Virginia, U.S.A., April 1997.

  • 7. N. Deshmukh, R. Duncan, and J. Picone, “Human Listening Benchmarks on

ARPA’s CSR Performance Tasks,” Proceedings Fourth International Conference

  • n

Spoken Language Processing, Philadelphia, Pennsylvania, U.S.A.,

  • pp. SuP1P1.10, October 1996.
  • 8. N. Deshmukh, M. Weber, and J. Picone, “Automated Generation of N-Best Pro-

nunciations of Proper Nouns,” Proceedings IEEE International Conference on Acoustics, Speech, and Signal Processing, Atlanta, Georgia, vol. 1, pp. 283-286, May 1996.

  • 9. N. Deshmukh and J. Picone, “Human Performance on ARPA’s CSR’95 Hub,” pre-

sented at the ARPA Spoken Language Systems Technology Workshop, Harriman, New York, January 1996. 10.W.J. Ebel and J. Picone, “Human Speech Recognition Performance on the 1994 CSR Spoke 10 Corpus,” Proceedings of the Spoken Language Systems Technol-

  • gy Workshop, pp. 53-59, Austin, Texas, January 1995.

11.Y. Muthusamy, E. Holliman, B. Wheatley, J. Picone, and J. Godfrey, “Voice Across Hispanic America: A Telephone Speech Corpus of American Spanish,” IEEE International Conference

  • n

Acoustics, Speech, and Signal Processing,

  • pp. 85-88, Detroit, Michigan, May 1995.