lecture 15 new
TRANSCRIPT
-
8/2/2019 Lecture 15 New
1/21
eturn to Main
jectives
sic Theory:
Simple Case
General Case
ovariance
Autocorrelation
rror
ectral Matching:
pectrum
Noise-Weightingreemphasis
Typical Front End
-Line Resources:
Signal Modeling
Linear Prediction
MAD SpeechL
E
C
T
U
R
E
15
:
LI
N
E
A
R
PREDICTION
0 Objectives:
0 Introduce the theory of linear prediction
0 Develop autocorrelation and covariance techniques for solution
0 Understand similarities with regression
0 Explain the relationship to windowing and maximum entropy
0 Add a new technique to our signal modeling block diagram
There is a classic textbook on this subject:
J.D. Markel and A.H. Gray, Linear Prediction of Speech , Springer-Verlag, New York,New York, USA, ISBN: 0-13-007444-6, 1976.
This lecture also includes material from two other textbooks:
J. Deller, et. al., Discrete-Time Processing of Speech Signals , MacMillan Publishing Co.,ISBN: 0-7803-5386-2, 2000.
and,
L.R. Rabiner and B.W. Juang, Fundamentals of Speech Recognition , Prentice-Hall, UppSaddle River, New Jersey, USA, ISBN: 0-13-015157-2, 1993.
-
8/2/2019 Lecture 15 New
2/21
turn to Main
oduction:
1: Organizatio
n ( html ,
pdf )
ech Signals:
2: Productio
n ( html ,
pdf )
3: Digital
Models
( html , pdf )
4: Perceptio
n ( html ,
pdf )
5: Maskin
g ( html ,
pdf )
6: Phonetics andPhonology ( html ,pdf )
7: Syntax andSemantics ( html ,pdf )
nal Processing:
8: Samplin
g ( html ,
pdf )
9: Resamplin
g ( html ,
pdf )
10: AcousticTransducers( html , pdf )
11: Temporal Analysis ( html , pdf )
12: Frequency Domain Analysis ( html , pdf )
13: Cepstral Analysis ( html , pdf )
14: Exam No. 1
( html , pdf )
15: Linear Prediction ( html , pdf )
16: LP-Based Representations ( html , pdf )
Parameterization:
17: Differentiation ( html , pdf )
18: Principal Components ( html , pdf )
ECE 8463: FUNDAMENTALS OF SPEECH
RECOGNITIONProfessor Joseph Picone
Department of Electrical and Computer Engineering
Mississippi State University
email: [email protected] phone/fax:601-325-3149; office: 413 Simrall
URL: http://www.isip.msstate.edu/resources/courses/ece_8463
Modern speech understanding systems merge interdisciplinary technologies from Signal Processing, PatternRecognition, Natural Language, and Linguistics into a unified statistical framework. These systems, which
have applications in a wide range of signal processing problems, represent a revolution in Digital Signal
Processing (DSP). Once a field dominated by vector-oriented processors and linear algebra-based mathematics,the current generation of DSP-based systems rely on sophisticated statistical models implemented using a
complex software paradigm. Such systems are now capable of understanding continuous speech input forvocabularies of hundreds of thousands of words in operational environments.
In this course, we will explore the core components of modern statistically-based speech recognition
systems. We will view speech recognition problem in terms of three tasks: signal modeling, network searching, and language understanding. We will conclude our discussion with an overview of state-of-the-art
systems, and a review of available resources to support further research and technology development.
http://www.isip.msstate.edu/publications/courses/ece_8463/http://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_01/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_01/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_01/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_01/lecture_01.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_01/lecture_01.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_01/lecture_01.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_02/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_02/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_02/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_02/lecture_02.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_02/lecture_02.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_03/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_03/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_03/lecture_03.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_03/lecture_03.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_04/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_04/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_04/lecture_04.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_04/lecture_04.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_05/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_05/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_05/lecture_05.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_05/lecture_05.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_06/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_06/lecture_06.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_06/lecture_06.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_07/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_07/lecture_07.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_07/lecture_07.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_08/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_08/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_08/lecture_08.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_08/lecture_08.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_08/lecture_08.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_09/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_09/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_09/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_09/lecture_09.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_09/lecture_09.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_10/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_10/lecture_10.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_11/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_11/lecture_11.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_12/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_12/lecture_12.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_13/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_13/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_13/lecture_13.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_13/lecture_13.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_14/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_14/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_14/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_14/lecture_14.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_14/lecture_14.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_15/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_15/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_15/lecture_15.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_15/lecture_15.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_16/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_16/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_16/lecture_16.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_16/lecture_16.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_17/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_17/lecture_17.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_17/lecture_17.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_18/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_18/lecture_18.pdfhttp://www.isip.msstate.edu/resources/courses/ece_8463http://www.isip.msstate.edu/resources/courses/ece_8463http://www.isip.msstate.edu/publications/courses/ece_8463/http://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_01/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_01/lecture_01.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_01/lecture_01.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_02/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_02/lecture_02.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_02/lecture_02.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_03/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_03/lecture_03.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_04/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_04/lecture_04.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_04/lecture_04.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_05/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_05/lecture_05.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_05/lecture_05.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_06/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_06/lecture_06.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_06/lecture_06.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_07/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_07/lecture_07.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_07/lecture_07.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_08/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_08/lecture_08.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_08/lecture_08.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_09/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_09/lecture_09.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_09/lecture_09.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_10/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_10/lecture_10.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_11/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_11/lecture_11.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_12/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_12/lecture_12.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_13/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_13/lecture_13.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_14/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_14/lecture_14.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_15/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_15/lecture_15.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_16/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_16/lecture_16.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_17/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_17/lecture_17.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_18/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_18/lecture_18.pdfhttp://www.isip.msstate.edu/resources/courses/ece_8463 -
8/2/2019 Lecture 15 New
3/21
files containing a compilationall the notes are available.
wever, these files are large
will require a substantialount of time to download. A
file of the html version of thees is available here . These
e generated using wget:
wget -np -k -mhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current
df file containing the entire setecture notes is available here .se were generated using Adobe
obat.
estions or comments about the
erial presented here can bected to [email protected] .
http://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_all/notes.html.tar.gzhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_all/notes.pdf.tar.gzmailto:[email protected]://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_all/notes.html.tar.gzhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_all/notes.pdf.tar.gzmailto:[email protected] -
8/2/2019 Lecture 15 New
4/21
19: Linear DiscriminantAnalysis ( html , pdf )
http://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_19/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_19/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_19/lecture_19.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_19/lecture_19.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_19/lecture_19.pdfhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_19/index.htmlhttp://www.isip.msstate.edu/publications/courses/ece_8463/lectures/current/lecture_19/lecture_19.pdf -
8/2/2019 Lecture 15 New
5/21
LECTURE 15: LINEAR PREDICTION
0 Objectives:
0 Introduce the theory of linear prediction
0 Develop autocorrelation and covariance techniques for solution
0 Understand similarities with regression
0 Explain the relationship to windowing and maximum entropy
0 Add a new technique to our signal modeling block diagram
ere is a classic textbook on this subject:
J.D. Markel and A.H. Gray, Linear Prediction of Speech , Springer-Verlag, New York, New York, USA, ISBN:0-13-007444-6, 1976.
is lecture also includes material from two other textbooks:
J. Deller, et. al., Discrete-Time Processing of Speech Signals , MacMillan Publishing Co., ISBN: 0-7803-5386-2,2000.
d,
L.R. Rabiner and B.W. Juang, Fundamentals of Speech Recognition , Prentice-Hall, Upper Saddle River,New Jersey, USA, ISBN: 0-13-015157-2, 1993.
-
8/2/2019 Lecture 15 New
6/21
LINEAR PREDICTION - A SIMPLE DERIVATION
-
8/2/2019 Lecture 15 New
7/21
LINEAR PREDICTION - GENERAL CASE
-
8/2/2019 Lecture 15 New
8/21
THE COVARIANCE METHOD
-
8/2/2019 Lecture 15 New
9/21
THE AUTOCORRELATION METHOD
-
8/2/2019 Lecture 15 New
10/21
LINEAR PREDICTION ERROR
-
8/2/2019 Lecture 15 New
11/21
SPECTRAL MATCHING INTERPRETATION
-
8/2/2019 Lecture 15 New
12/21
NOISE FLOORING
VIA THE AUTOCORRELATION FUNCTION
-
8/2/2019 Lecture 15 New
13/21
SIGNAL CONDITIONING COMPENSATES FOR
MICROPHONE AND CHANNEL CHARACTERISTICS
-
8/2/2019 Lecture 15 New
14/21
A TYPICAL SPEECH RECOGNITION FRONT END
-
8/2/2019 Lecture 15 New
15/21
ndex of ublications/journals/ieee_proceedings/1993/signal_modeling
Name Last modified SizeDescription
Parent Directory01-Jan-1999 12:07 -
paper_v2.pdf22-Jun-1999 16:14 452k
ache/1.3.9 Server at www.isip.msstate.edu Port 80
http://www.isip.msstate.edu/publications/journals/ieee_proceedings/1993/signal_modeling/?M=Ahttp://www.isip.msstate.edu/publications/journals/ieee_proceedings/1993/http://www.isip.msstate.edu/publications/journals/ieee_proceedings/1993/signal_modeling/paper_v2.pdfhttp://www.isip.msstate.edu/publications/journals/ieee_proceedings/1993/signal_modeling/?M=Ahttp://www.isip.msstate.edu/publications/journals/ieee_proceedings/1993/http://www.isip.msstate.edu/publications/journals/ieee_proceedings/1993/signal_modeling/paper_v2.pdf -
8/2/2019 Lecture 15 New
16/21
xt: Motivation from lossless tubes Up: Speech Analysis Previous: Z transforms
inear Prediction analysis
near prediction analysis of speech is historically one of the most important speech analysis techniques. The basis issource-filter model where the filter is constrained to be an all-pole linear filter. This amounts to performing a linear
ediction of the next sample as a weighted sum of past samples:
is linear filter has the transfer function:
good introductory article is [ 8], and this subject is also covered well in [ 1, 2, 3].
0 Motivation from lossless tubes0 Parameter estimation0 The autocorrelation method0 The covariance method0 Pre-emphasis0 The LP spectrum0 Gain computation0 The lattice filter implementation0 The Itakura distance measure0 The LP cepstrum0 Log area ratios0 The roots of the predictor polynomial0 Line spectral pairs
peech Vision Robotics group / Tony Robinson
http://svr-www.eng.cam.ac.uk/~ajr/SA95/node39.htmlhttp://svr-www.eng.cam.ac.uk/~ajr/SA95/SpeechAnalysis.htmlhttp://svr-www.eng.cam.ac.uk/~ajr/SA95/node37.htmlhttp://svr-www.eng.cam.ac.uk/~ajr/SA95/node93.html#Makhoul75http://svr-www.eng.cam.ac.uk/~ajr/SA95/node93.html#RabinerSchafer78http://svr-www.eng.cam.ac.uk/~ajr/SA95/node93.html#FallsideWoods85http://svr-www.eng.cam.ac.uk/~ajr/SA95/node93.html#FallsideWoods85http://svr-www.eng.cam.ac.uk/~ajr/SA95/node93.html#DellerProakisHansen93http://svr-www.eng.cam.ac.uk/~ajr/SA95/node93.html#DellerProakisHansen93http://svr-www.eng.cam.ac.uk/~ajr/SA95/node39.html#SECTION00071000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node40.html#SECTION00072000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node41.html#SECTION00073000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node42.html#SECTION00074000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node43.html#SECTION00075000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node44.html#SECTION00076000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node45.html#SECTION00077000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node46.html#SECTION00078000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node47.html#SECTION00079000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node48.html#SECTION000710000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node49.html#SECTION000711000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node50.html#SECTION000712000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node51.html#SECTION000713000000000000000http://svr-www.eng.cam.ac.uk/http://svr-www.eng.cam.ac.uk/~ajrhttp://svr-www.eng.cam.ac.uk/~ajr/SA95/node39.htmlhttp://svr-www.eng.cam.ac.uk/~ajr/SA95/SpeechAnalysis.htmlhttp://svr-www.eng.cam.ac.uk/~ajr/SA95/node37.htmlhttp://svr-www.eng.cam.ac.uk/~ajr/SA95/node93.html#Makhoul75http://svr-www.eng.cam.ac.uk/~ajr/SA95/node93.html#RabinerSchafer78http://svr-www.eng.cam.ac.uk/~ajr/SA95/node93.html#FallsideWoods85http://svr-www.eng.cam.ac.uk/~ajr/SA95/node93.html#DellerProakisHansen93http://svr-www.eng.cam.ac.uk/~ajr/SA95/node39.html#SECTION00071000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node40.html#SECTION00072000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node41.html#SECTION00073000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node42.html#SECTION00074000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node43.html#SECTION00075000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node44.html#SECTION00076000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node45.html#SECTION00077000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node46.html#SECTION00078000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node47.html#SECTION00079000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node48.html#SECTION000710000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node49.html#SECTION000711000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node50.html#SECTION000712000000000000000http://svr-www.eng.cam.ac.uk/~ajr/SA95/node51.html#SECTION000713000000000000000http://svr-www.eng.cam.ac.uk/http://svr-www.eng.cam.ac.uk/~ajr -
8/2/2019 Lecture 15 New
17/21
linear prediction of speech
introduction | demonstration | investigate | reading | credits | downloading | home
ntroduction
This tool allows users to explore linear prediction of speech and other signals. It also reinforces theconcepts of windowing, preemphasis and frame size selection in speech analysis. Users can selectregions for analysis and see the results as lpc-smoothed spectra and poles. The residual signal and itsspectrum is also displayed. This tool is not a substitute for learning about linear prediction. Tounderstand what the various panels mean, you should read one of the texts listed below. The demorequires the MATLAB signal processing toolbox.
he tool
http://svr-www.eng.cam.ac.uk/~ajrhttp://svr-www.eng.cam.ac.uk/~ajrhttp://svr-www.eng.cam.ac.uk/~ajrhttp://svr-www.eng.cam.ac.uk/~ajrhttp://svr-www.eng.cam.ac.uk/~ajrhttp://svr-www.eng.cam.ac.uk/~ajrhttp://svr-www.eng.cam.ac.uk/~ajrhttp://svr-www.eng.cam.ac.uk/~ajrhttp://svr-www.eng.cam.ac.uk/~ajrhttp://svr-www.eng.cam.ac.uk/~ajrhttp://svr-www.eng.cam.ac.uk/~ajrhttp://svr-www.eng.cam.ac.uk/~ajrhttp://www.dcs.shef.ac.uk/~martin/MAD/docs/download.htmhttp://www.dcs.shef.ac.uk/~martin/MAD/docs/mad.htmhttp://www.dcs.shef.ac.uk/~martin/MAD/docs/mad.htmhttp://www.dcs.shef.ac.uk/~martin/MAD/docs/download.htmhttp://www.dcs.shef.ac.uk/~martin/MAD/docs/mad.htm -
8/2/2019 Lecture 15 New
18/21
Type 'lpcspect' to launch the demo. The file menu is used to load an existing signal file or to create a new one
using the createsig tool. Supported formats currently include .wav, .snd and .au sound files. The signal
will appear in panel (1), and associated linear prediction signals for the region between the cursors will bedisplayed in the other panels. Use the cursors to select any segment. On cursor release, the segmentappears in panel (2) along with linear prediction residual (error signal). Panel (3) shows the DFTspectrum of this segment, overlaid with the LPC-smoothed spectrum and the DFT of the residual (7). Thepoles corresponding to the linear predictor are shown in panel (4).
The LPC order can be changed using menu (8). The signal can be preemphasised by checking therelevant box in cluster (6). Cursors can be linked (so that a constant segment length is maintained). Thewaveform segment chosen will be played on release if the associated checkbox is checked.
Clicking on any signal waveform (incuding the residual) causes it to be played. Holding the mousebutton down in the z-plane panel (4) will cause the frequency at the selected z-plane angle to be
http://www.dcs.shef.ac.uk/~martin/MAD/library/createsig.htmhttp://www.dcs.shef.ac.uk/~martin/MAD/library/createsig.htm -
8/2/2019 Lecture 15 New
19/21
displayed in the top left corner of the panel.
The chosen waveform segment can be Hamming-windowed (5). Unchecked, a rectangular window isapplied.
-
8/2/2019 Lecture 15 New
20/21
hings to investigate
1. Choose a vowel segment of speech. Examine what happens to the residual signal (and its meansquare error) as you increase the number of poles. What happens to the LPC-smoothedspectrum? How many poles provide a comfortable fit to a vowel segment.
2. Now repeat this analysis for unvoiced sounds such as fricatives. How does the error compare?3. Again, perform this analysis for a nasal sound.4. Can you predict where the poles will appear for a given analysis order, and how they will
change when you modify the order?5. Explore the effect of preemphasis on the LPC spectrum and the number of poles required
to obtain a given residual error.6. Listen to the waveform segment and the residual signal in sequence. Can you detect a change
in timbre? How does this change manifest itself as the LPC order increases?7. What is the effect of changing LPC order on the spectrum of the residual signal?8. Choose a vowel sound and a relatively low LPC order (10, say). Read off the pole locations
(frequencies) by holding the mouse down in panel (4). Now use the polezero tool to recreate
just these pole locations and listen to the resulting sound.
9. Use the create option on the file menu to create a harmonic series with, say, 5 components.Do you expect to be able to fit it well with a 10th order model? Try it.
10. Now add some noise (using create ) to your harmonic series. Do the pole locations change? (Readabout the peak-hugging properties of LPC).
urther reading
0 A good introduction to LPC is provided by Makhoul, J. (19??)0 This tool was written partly as an example of the interface issues in speech and
hearing demonstrations, and is described further in Cooke et al (1999) The interactiveauditory demonstrations project, Eurospeech'99, Budapest, September.
redits etc
Produced by : Martin Cooke, April 1999.
Permissions : This demonstration may be used and modified freely by anyone. It may be distributed
in unmodified form.
http://www.dcs.shef.ac.uk/~martin/MAD/polezero/polezero.htmhttp://www.dcs.shef.ac.uk/~martin/MAD/polezero/polezero.htm -
8/2/2019 Lecture 15 New
21/21
F to Word
http://pdfonline.blogspot.com/http://pdfonline.blogspot.com/