Robust Speech Recognition and its ROBOT implementation

All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Robust Speech Recognition and
its ROBOT implementation
Yoshikazu Miyanaga
Hokkaido University
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Conditions for Speech Recognition
Short Isolated Speech:
words, phrase (<2sec)
Continuous Speech:
sentences
(>2sec)
Attached
Mic
(several cm
–
10cm)
Remote
Mic
:
(10cm
–
5m)
Silent Room
（
>20dB)
Living Room
（
20
～
10dB)
Noisy Room:
exhibition
（
<1
0dB)
Long Distance
Mic
:
(>5m)
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Conventional ASR
Continuous
Speech
:
(>2sec)
Attached
Mic
(<10cm)
Silent Room
（
>20dB)
Attached
Mic
(<10cm)
Short Isolated
Speech
:
(<2sec)
Living Room
（
20
～
10dB)
Short Isolated
Speech: (<2sec)
Attached Remote
Mic
:
(<5m)
Living Room
（
20
～
10dB)
Array Microphone
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Hokkaido University Speech
Communication System (HU
-
SCS)
Short Isolated Speech:
words, phrase (<2sec)
Remote
Mic
:
(10cm
–
5m)
Long Distance
Mic
:
(>5m)
Silent Room
（
>20dB)
Living Room
（
20
～
10dB)
Noisy Room:
exhibition
（
<10dB)
Attached
Mic
(several cm
–
10cm)
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
HU
-
SCS
Automatic
Speech Detection
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
HU
-
SCS
Automatic
Speech Detection
97% by Current Technology
（
SNR 10dB)
・
WAVELET
・
Non
-
Linear Processing
“Robust voice activity detection using
perceptual wavelet
-
packet transform and
teager
energy operator” S
-
H Chen, H
-
T Wu,
Y. Chang and T.K. Truong, Trans. Pattern
Recognition Letters (2007)
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
HU
-
SCS
Automatic
Speech Detection
HU
-
SCS
v4
・
99% over SNR 10dB
・
BP
＋
Threshold
Ope
・
F
0
Detection
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
HU
-
SCS
Automatic
Speech
Recognition
Candidates of Recognition Results
(1)
Good Morning
(2)
See you
(3)
How are you ?
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
HU
-
SCS
Automatic
Speech
Recognition
Candidates of Recognition Results
(1)
Good Morning
(2)
See you
(3)
How are you ?
71%
by Current Tech
（
SNR 10dB) .
97.4%
(SNR 20dB).
・
Spectral Subtraction
・
RASTA, CMS
・
A Prior Information
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
HU
-
SCS
Automatic
Speech
Recognition
Candidates of Recognition Results
(1)
Good Morning
(2)
See you
(3)
How are you ?
HU
-
SCS v4
・
95.3% (
SNR
10dB).
98.3% (20dB)
・
No A Prior Info.
・
RSF/DRA
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
HU
-
SCS
Automatic
Speech Rejection
Recognition Result
Good
Morning
Candidates of Recognition Results
(1)
Good Morning
(2)
See you
(3)
How are you ?
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
HU
-
SCS
Automatic
Speech Rejection
Recognition Result
Good
Morning
Candidates of Recognition Results
(1)
Good Morning
(2)
See you
(3)
How are you ?
90%
by Current Tech
・
Confidential Scoring
Technique
“Recognition confidential
scoring and its use in speech
understanding systems”, T.J.
Hazen,
S.Seneff
and
J.Polifroni
, Trans on Computer
Speech and language (2002).
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
HU
-
SCS
Automatic
Speech Rejection
Recognition Result
Good
Morning
Candidates of Recognition Results
(1)
Good Morning
(2)
See you
(3)
How are you ?
HU
-
SCS v4
・
Dependent GMM by
Weighted
HMM
(90%
Accuracy)
・
AI (Artificial
Intelligence)
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
HU
-
SCS
Automatic
Speech Detection
Automatic
Speech
Recognition
Automatic
Speech Rejection
HW with
Low Power
・
Super Low
-
Power Consumption
Design
・
Real
-
Time SCS
→
180nsec/word (10MHz
クロック）
Recognition Time
・
Small Scale Design with Special Designed LSI
・
Noise Reduction by Array Microphone
First
SCS
HW
・
LSI IP
・
Mobile
・
Intelligent Consumer Electronics
etc
Fine Advantage
(1)
Mobile
Appli
•
Small
•
Low
Power
(2)
PC free
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
HU
-
SCS
Automatic
Speech Detection
Automatic
Speech
Recognition
Automatic
Speech Rejection
HW with
Low Power
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Running Spectrum Domain
Waveform
・・・
Mel
-
Spectra
1 2 3
・・・
t
・・・
1 2 3
・・・
t
-
6
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
BP and Threshold OP
Start Point
End Point
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Detection
Recognition
Operation/Control
Operation/Control
Recognition
Operation/Control
Switch
-
Less Recognition System by Automatic Detection
無音区間
Start
Speech
無音区間
End
Recognition
Start
Recognition
End
“
Hands Free
”
剥R潧o楴楯i
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
HU
-
SCS
Automatic
Speech Detection
Automatic
Speech
Recognition
Automatic
Speech Rejection
HW with
Low Power
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Speech Analysis and Robust Processing

Speech Analysis

LPC Cepstrum

Mel
-
Frequency Cepstrum

Robust Processing

Various types of techniques have been proposed.

Spectral Subtraction

Wiener Filtering

Microphone Arrays

RSF/DRA (Running Spectrum Filtering/Dynamic
Range Adjustment)

uses filtering and normalizing for cepstral vectors.
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Procedure of Mel
-
Frequency Cepstrum
Speech Signals
Cut into Short
-
Time Frames
Discrete Fourier Transform (DFT)
Filterbanks with Mel
-
Frequency Scale
Logarithm
Discrete Cosine Transform (DCT)
x(t)
x
f
(n,t
s
)
|X(n,f)|
X
s
(n,f
m
)
log(X
s
(n,f
m
))
C(n,k)
Cepstral Coefficients
n : frame index
k : cepstral order
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Noise modeling

Spectrum including noise can be modeled as,
)
(
)
(
)
,
(
)
,
(




A
H
n
S
n
X


Clean
spectrum
Multiplicative
noise
Additive noise
)
(
)
,
(
)
,
(



H
n
S
n
E

log
))
(
)
,
(
log(
)
,
(
log



A
n
E
n
X


All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Noise Corruption in Power Spectrum
E(n,
ω)+A
E(n,
ω)

Noise corruptions make differences on
gains and DC components.
Clean Speech
Noisy Speech
Power Spectrum
(White Noise
at 10dB
SNR)
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Noise Corruption in Log Power Spectrum
E(n,
ω)+A
E(n,
ω)

Noise corruptions make differences on
gains and DC components.
DC Components
Gain
Log
-
power Spectrum
Clean Speech
Noisy Speech
(White Noise
at 10dB
SNR)
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Running Spectrum

Running spectrum is obtained by accumulating
short
-
time spectrum
0
2000
4000
6000
8000
10000
12000
-0.1
0
0.1
0.2
DFT
Frame Number
Frequency
Running spectrum: time trajectory
of frequency
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Spectral Subtraction
Estimate the spectrum of noise
from short
-
time spectra in the
first several flames
Running spectrum of a noisy speech
(white noise at 5 dB SNR)
Subtract the estimated
spectrum from each
short
-
time spectrum
After Subtraction
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Noise Reduction Techniques

Conventional method

Spectral subtraction

Parameters are not optimized for speeches from various
environments.

Excessive subtraction may cause musical noise.

Robust speech feature extraction.

Advanced speech analysis using RSF (running
spectral filtering) and DRA (dynamic range
adjustment).
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Modulation Spectrum
Modulation Spectrum
Running Spectrum
Frame Number
Frequency
DFT on each frequency
Frequency
Modulation
frequency

RSF focuses on modulation spectrum
Modulation spectrum: spectrum versus time
trajectory of frequency.
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Mod
-
F of Clean and Noisy Speech
Clean
Noisy (white noise at 5 dB SNR)

Speech components are dominant around
4 Hz in modulation spectrum.
Lower modulation frequency components can be assumed as
noise because of little changes in noise components.
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
RSF (Running Spectrum Filtering)

Speech components are
dominant around
4 Hz in modulation spectrum.
Modulation Frequency [Hz]
Modulation Spectrum
Noise Components
Speech Components
Unnecessary Part
Frequency (Hz)
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
RSF

RSF (Running Spectrum Filtering)

enhances perceptual auditory components.

decreases noise components relatively by band
-
pass filtering in cepstral sequences.
)
,
(
)
(
)
,
(
~
0
k
i
n
C
i
h
k
n
C
Q
i





Coefficients in FIR Filter
Modulation
Frequency
of RSF
RSF
RASTA(IIR)
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
DRA

DRA (Dynamic Range Adjustment)

normalizes amplitude of cepstral vectors in time
domain (use of maximum value during utterance).

suppresses dynamic range distortions caused by
additive noise.
k
k
n
C
k
n
C

)
,
(
~
)
,
(

|
)
,
(
~
|
max
1
k
n
C
T
k
k




All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
RSF / DRA
10
20
30
40
50
60
70
80
90
100
-3
-2
-1
0
1
2
3
RSF processing
10
20
30
40
50
60
70
80
90
100
-3
-2
-1
0
1
2
Baseline
10
20
30
40
50
60
70
80
90
100
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
RSF/DRA processing
Clean
Noisy

Comparison in cepstral time
-
trajectories at 4th order
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
HU
-
SCS
Automatic
Speech Detection
Automatic
Speech
Recognition
Automatic
Speech Rejection
HW with
Low Power
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Likelihoods of HMM
HMM
GMM
GMM
GMM
GMM
GMM
Approximation of many multi
-
dimensional Gaussian
Distribution
Average
Variance
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Evaluation on Likelihoods
MFCC
Likelihood of MFCC
into this HMM
1
p
2
p
4
p
3
p
5
p
6
p
7
p
8
p
9
p
11
p
10
p
The maximum likelihood
is selected and its label is
recognized as the result.
The result is
correct, isn’t it ?
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Evaluation of Reliability
The result of the top
score is trusted.
Likelihood
Likelihood
The result of the top
score is
NOT
trusted.
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Rejection Method using Multi
-
Criterions
Tendency
Maximum Score
Ratio
MFCC Cluster
Square of Ratio
Group Evaluation of Cluster
Noisy Conditions
New Type
Speech
Rejection
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
HU
-
SCS
Automatic
Speech Detection
Automatic
Speech
Recognition
HW with
Low Power
Automatic
Speech Rejection
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Overview of ASR System

Current ASR systems adopt robust processing
that removes influences of noise distortions.
Speech
Data
Speech
Analysis
Covert to Spectrum or Cepstrum
Robust
Processing
Decrease Noise Distortions
Speech
Recognition
Calculate Probability (likelihood
scores)
Results
Reference Models
Prepare Reference Patterns by Speech Training
Speech Feature
Vectors
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Circuit Structure of
Complete Recognition System
System Control
16
16
16
Robust
Processing
SRAM
16
16
16
16
16
16
Speech
Recognition
Speech
Analysis
24
24
SRAM
SRAM
24
24
16
16
Data
Control
External
Memory
(SRAM)
16
Speech Signal
from/to
Processor
16
16
System Control
16
16
16
Robust
Processing
SRAM
16
16
16
16
16
16
Speech
Recognition
Speech
Analysis
24
24
SRAM
SRAM
24
24
16
16
Data
Control
External
Memory
(SRAM)
16
Speech Signal
from/to
Processor
16
16
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Circuit Implementations

Required Operating Performance

Speech Analysis
…
10 MIPS

Robust Processing
…
500 MIPS (mainly in FIR)
Buffer
256*16
bits
16
FFT
IDCT
Log
Cos/Sin
ROM
RAM
512*24
bits
Buffer
4096*16
bits
FIR
Divider
ROM
16
16
16
8
Speech Analysis
(MFCC)
Robust Processing
(RSF/DRA)
Speech
Data
Feature
Vectors
RAM
256*24
bits
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Block Diagram

Interfaces

Microprocessor, External RAM, and Master/Slave
MPU Interface
HMM
16
Master Bus
16
5
5
24
RSF/DRA
24
16
16
MFCC
24
24
SRAM
16
16
Bus Control
System Control
SRAM
interface
16
2
1
20
Address
Interrupt Signal
Chip Select
16
16
SRAM
24
24
SRAM
16
Filter Coefficients for RSF
Working for MFCC and RSF
Feature parameters before speech detection
16
16
1
2
2
2
Slave Bus
Data Control
3
Data Control
5
SW
CLK
RESET
MPU Interface
HMM
16
Master Bus
16
5
5
24
RSF/DRA
24
16
16
MFCC
24
24
SRAM
16
16
Bus Control
System Control
SRAM
interface
16
2
1
20
Address
Interrupt Signal
Chip Select
16
16
SRAM
24
24
SRAM
16
Filter Coefficients for RSF
Working for MFCC and RSF
Feature parameters before speech detection
16
16
1
2
2
2
Slave Bus
Data Control
3
Data Control
5
SW
CLK
RESET
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
New Scalable Architectures

2 types of scalable techniques are applied to the
system.
(1)
Multiple Process Elements (PEs) in HMM Circuit
The PEs enable high
-
speed processing and improving
recognition performance.
(2)
Master/Slave Operation in the Complete System
The operation enables high
-
speed processing and
increase the number of word vocabularies.
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
HMM (Hidden Markov Models)

Hidden Markov Models (HMM)

Statistical modeling approach using Markov chain.

Powerful for expressing time
-
varying data sequences
and robust with speaker differences.
11
a
22
a
12
a
44
a
33
a
34
a
23
a
45
a
1
q
2
q
3
q
4
q
ij
a
State transition probability
)
1
(
N
n
n
q


Set of states
)
(
1
k
b
)
(
2
k
b
)
(
k
b
N
Output probability
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Full
-
Parallel
Computations in HMM

The output probabilities and temporal scores can be
computed concurrently for the number of HMM states.
Output Prob. Calc.
Output Prob. Calc.
Output Prob. Calc.
Output Prob. Calc.
o
t
Score Calc.
Score Calc.
Score Calc.
Score Calc.
Path for upper state
Select
Max
Max(
δ
)
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Microprocessor
RAM
Master
Slave1
Slave2
Slave3
Master/Slave
Operation
(1)
Set Reference Data
(2)
Speech Analysis and
Robust Processing
(3)
Broadcast
(4)
Speech Recognition
(5)
Gather Results
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Microprocessor
RAM
Master
Slave1
Slave2
Slave3
Master/Slave
Operation
(1)
Set Reference Data
(2)
Speech Analysis and
Robust Processing
(3)
Broadcast
(4)
Speech Recognition
(5)
Gather Results
[1]
[2]
[3]
[4]
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Microprocessor
RAM
Master
Slave1
Slave2
Slave3
Master/Slave
Operation
(1)
Set Reference Data
(2)
Speech Analysis and
Robust Processing
(3)
Broadcast
(4)
Speech Recognition
(5)
Gather Results
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Microprocessor
RAM
Master
Slave1
Slave2
Slave3
Master/Slave
Operation
(1)
Set Reference Data
(2)
Speech Analysis and
Robust Processing
(3)
Broadcast
(4)
Speech Recognition
(5)
Gather Results
[2]
[1]
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Microprocessor
RAM
Master
Slave1
Slave2
Slave3
Master/Slave Operation(2)
(1)
Set Reference Data
(2)
Speech Analysis and
Robust Processing
(3)
Broadcast
(4)
Speech Recognition
(5)
Gather Results
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Microprocessor
RAM
Master
Slave1
Slave2
Slave3
Master/Slave Operation(2)
(1)
Set Reference Data
(2)
Speech Analysis and
Robust Processing
(3)
Broadcast
(4)
Speech Recognition
(5)
Gather Results
[1]
[2]
[3]
[4]
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Circuit Design (Analysis & HMM TEG)

Technology

Rohm
CMOS 0.35
μ
m

Univ. of Tokyo EXD
Standard
Cell Library

Voltage Supply
3.3V

RTL
Level Design
……
.
Verilog
-
HDL

Evaluation
Clock Freq.
(MHz)
Proc Time
(ms/word)
Power
Coms
(
mW
)
60
0.029
567.7
30
0.059
285.2
10
0.180
93.2
V2 Layout View
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Comparison on Power Consumption

Proposed HW (10MHz) and DSP Design (80MIPS)
DSP based System
Proposed System
Processor Structure
TMS320C549
80MIPS
Dedicated
Processor
10MHz
Memory Access
Time (ns)
15
80
Processor (mW)
(Core : 3.3V)
158.4
93.2
Memory (mW)
(SRAM, Core : 3.3V)
627
100
Total
785.4
193.2
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Processing Time of HU
-
SCS

Comparison with Software Design

54 times faster

No high speed clock
……
Useful for Low
-
Power Design
Proposed System
(Hardware)
Pentium 4
(Software)
No. arithmetic units
160
-
No. cycles
455,200
-
Frequency(MHz)
80
2200
Recognition
Processing time(ms)
5.7
310
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Design by Standard Cells

TSMC0.25
µ
m CMOS
Standard Cell

Voltage 2.5V

Highest Clock Rate 80.6MHz
(12.4ns,
Temperature Cond. Typical
)
No. Parallel Processing
32 8
HMM
491,600 116,980
RSF/DRA
11,910
MFCC
39,670
System Control
18,310
Bus Control
1,310
SRAM
63,400
Total
626,200 251,580
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Current HU
-
SCS
HU
-
SCS Board
PC
Interface with
HU
-
SCS Board
55mm
×
44 mm
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Overview of Current HU
-
SCS

Improvement of Noise Robust

Accurate ASR under SNR
0
-
10dB

Robustness against Echo

Improvement of Speech Recognition

Higher Accuracy on MFCC Calculation

Low Power Design and Higher Speed
Processing

Improvement of Total HW System

Higher Speed Response Time
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Comparison on Performance
Environment
Noise Level
Correctness
Current
Previous
Meeting Room
50
ｄ
B
96.4%
90.0%
Elevator
50
ｄ
B
95.0%
84.4%
Stairs
45
ｄ
B
85.1%
50.5%
Car A
（
Idling, No
-
Moving
）
50
ｄ
B
99.4%
95.6%
Car B
（
High Speed,
Open Window
）
75
ｄ
B
93.3%
85.0%
Car C
（
High Speed,
Audio
ON
（ＦＭ））
75
ｄ
B
88.9%
65.6%
Total
93.0%
78.5%
※Cruiser Board
（
Outside,
high speed
）
80
ｄ
B
82.7%
-
Comparisons
between HU
-
SCS
v4 and v3
0.00%
50.00%
100.00%
Previous
Current
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Results on Some Distances
60.0%
70.0%
80.0%
90.0%
100.0%
30cm
60cm
90cm
Car A
30.0%
40.0%
50.0%
60.0%
70.0%
80.0%
90.0%
100.0%
30cm
60cm
90cm
Car C
60.0%
70.0%
80.0%
90.0%
100.0%
30cm
60cm
90cm
Elevator
30.0%
40.0%
50.0%
60.0%
70.0%
80.0%
90.0%
100.0%
30cm
60cm
90cm
Stair
60.0%
70.0%
80.0%
90.0%
100.0%
30cm
60cm
90cm
Meeting Room
60.0%
70.0%
80.0%
90.0%
100.0%
30cm
60cm
90cm
Car B
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Robot Implementation

Speech Recognition & Synthesis

Quick Response

Control to Consumer Electronics and
Machines




All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Communications and Controls
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Summary

Hokkaido University Speech Communication
System

Integrated Architecture of
Speech Detection, Robust
Speech
Analysis, Speech Recognition,
Speech Rejection

Higher
Speed Processing
than DSP and Software

Superior in Energy Saving than DSP Solutions

Improving Noise Robustness by RSF/DRA
Technique

Small, Fast and Low Power
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Who ?
64
Yoshikazu
Miyanaga
He
received the B.S., M.S., and Dr. Eng. degrees from Hokkaido University, Sapporo,
Japan, in 1979, 1981, and 1986, respectively. He is currently a Professor at Graduate
School of Information Science and Technology, Hokkaido University.
His research interests
are in the areas of signal processing for wireless
communications, nonlinear signal processing and low
-
power LSI systems.
He
was a chair of Technical Group on Smart Info
-
Media System, IEICE. He is an
advisory member of this technical group.
Currently, he is IEICE fellow.
He
served as a member in the board of directors, IEEE Japan Council as a chair of
student activity committee from 2002 to 2004. He is a chair of student activity
committee in IEEE Sapporo Section from 2001. He is a chair of IEEE Circuits and
Systems Society, Digital Signal Processing Technical Committee from 2006.
He
has been serving as international steering committee chairs/members of IEEE
ISPACS, IEEE ISCIT, IEEE/EURASIP NSIP and honorary/general chairs/co
-
chairs of their
international symposiums/workshops, i.e., ISPACS 2003, ISCIT 2004, ISCIT 2005, NSIP
2005, ISPACS 2008, ISMAC 2009 and APSIPA ASC 2009. He also served as
international organizing committee chairs of IEICE ITC
-
CSCC 2002
-
2003, IEEE MSCAS
2004, IEEE ISCAS 2005
-
2008.
All right reserved. Copyright ©
2009
-
Yoshikazu
Miyanaga
Current References of this Topic
1.
Kazunaga
Ohnuki
,
Wataru
Takahashi, Shingo
Yoshizawa
, Yoshikazu
Miyanaga
, “Noise Robust Speech Features for Automatic Continuous Speech
Recognition using Running Spectrum Analysis”, Proceedings of 2008 International Symposium on Communications and Information T
ech
nologies
(ISCIT), pp.150
-
153, October 2008.
2.
Jirabhorn
Chaiwongsai
,
Werapon
Chiracharit
,
Kosin
Chamnongthai
, Yoshikazu
Miyanaga
, “An Architecture of HMM
-
Based Isolated
-
Word Speech
Recognition with Tone Detection Function”, Proceedings of 2008 International Symposium on Intelligent Signal Processing and C
omm
unication Systems
(ISPACS), December 2008.
3.
Nongnuch
Suktangman
,
Kham
Khanthavivone
,
Kraisin
Songwatana
, Yoshikazu
Miyanaga
, “Robust Speech Recognition Based on Speech Spectrum on
Bark Scale”, EURASIP Proceedings of 2007 International Workshop on Nonlinear Signal and Image Processing (NSIP), pp.135
-
138, Sep
tember 2007.
4.
Shingo
Yoshizawa
,
Naoya
Wada, Noboru
Hayasaka
, Yoshikazu
Miyanaga
, "Scalable Architecture for Word HMM
-
Based Speech Recognition and VLSI
Implementation in Complete System", IEEE Transactions on Circuits and Systems I, Vol.53, No.1, pp.70
-
77, January 2006.
5.
Noboru
Hayasaka
and Yoshikazu
Miyanaga
, “Spectrum Filtering with FRM for Robust Speech Recognition”, IEEE Proceedings of International
Symposium on Circuits and Systems (ISCAS), No.2, pp.3285
-
3288, May 2006.
6.
Naoya
Wada, Noboru
Hayasaka
, Shingo
Yoshizawa
, Yoshikazu
Miyanaga
, “Direct Control on Modulation Spectrum for Noise
-
Robust Speech
Recognition and Spectral Subtraction,” IEEE International Symposium on Circuits and Systems (ISCAS), pp. 2533
-
2536, May 2006.
7.
Shingo
Yoshizawa
, Noboru
Hayasaka
,
Naoya
Wada, Yoshikazu
Miyanaga
, “VLSI Architecture for Robust Speech Recognition Systems and its
Implementation in Verification Platform,” Journal of Robotics and
Mechatronics
, Vol.17, No.4, pp. 447
-
455, Aug. 2005.
8.
Yasuyuki
Hatakawa
, Shingo
Yoshizawa
, Yoshikazu
Miyanaga
, “Robust VLSI Architecture for System
-
On
-
Chip Design and its implementation in
Viterbi
Decoder,” IEEE International Symposium on Circuits and Systems (ISCAS), Vol.3, pp.25
-
28, May 2005.
9.
K.Songwatana
, K.
Dejhan
, Y.
Miyanaga
and K.
Khanthavivone,“A
Vowels Recognition Model for
Laotion
language using Transfer Function on Bark
scale and Hidden Markov Modeling”, IEEE Proceedings of International Workshop on Nonlinear Signal and Image Processing (NSIP)
, V
ol.1, pp.426
-
429,
May 2005.
10.
Kazuma
Fujioka,Noboru
Hayasaka,Yoshikazu
Miyanaga
and
Norinobu
Yoshida,“A
Noise Reduction Method of Speech Signals Using Running Spectrum
Filtering”, IEICE Transactions on Information and Systems Part.2,Vol.J88
-
D
-
Ⅱ
, No.4,pp.695
-
703,April 2005.
11.
Qi
Zhu, Noriyuki
Ohtsuki
, Yoshikazu
Miyanaga
and
Norinobu
Yoshida,“Noise
-
Robust Speech Analysis Using Running Spectrum Filtering”, IEICE
Transactions on Fundamentals of Electronics, Communications and Computer Science, Vol.E
-
88
-
A, No.2, pp. 541
-
548, February 2005.
65