JP2003271168A - Method, device and program for extracting signal, and recording medium recorded with the program - Google Patents

Method, device and program for extracting signal, and recording medium recorded with the program

Info

Publication number
JP2003271168A
JP2003271168A JP2002072111A JP2002072111A JP2003271168A JP 2003271168 A JP2003271168 A JP 2003271168A JP 2002072111 A JP2002072111 A JP 2002072111A JP 2002072111 A JP2002072111 A JP 2002072111A JP 2003271168 A JP2003271168 A JP 2003271168A
Authority
JP
Japan
Prior art keywords
signal
band
sub
program
mixed signals
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
JP2002072111A
Other languages
Japanese (ja)
Inventor
Akiko Araki
Shoji Makino
Makoto Mukai
Hiroshi Saruwatari
Hiroshi Sawada
良 向井
宏 澤田
昭二 牧野
洋 猿渡
章子 荒木
Original Assignee
Nippon Telegr & Teleph Corp <Ntt>
日本電信電話株式会社
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nippon Telegr & Teleph Corp <Ntt>, 日本電信電話株式会社 filed Critical Nippon Telegr & Teleph Corp <Ntt>
Priority to JP2002072111A priority Critical patent/JP2003271168A/en
Publication of JP2003271168A publication Critical patent/JP2003271168A/en
Application status is Pending legal-status Critical

Links

Abstract

<P>PROBLEM TO BE SOLVED: To provide a signal extracting method, a signal extracting device and a signal extraction program for estimating a separation filter with a length sufficient for separation while holding the assumption of the independence a signal in each band by adopting a configuration for applying subband analysis combination to a plurality of mixed signals to calculate an output signal corresponding to the original signal, and to provide a recording medium recorded with the program. <P>SOLUTION: This signal extracting method is for separatingly extracting the original signal from the plurality of mixed signals observed through a path having a long impulse response on the basis of its independence. By using the signal extracting method, the signal extracting device, the signal extraction program and the recording medium recorded with the program, the plurality of mixed signals are inputted to a subband analyzing part 51, the mixed signals are respectively subjected to a subband analysis into N (N is positive integer) bands, the subband-analyzed signal of each band is inputted to a time domain BSS part 53 of a corresponding band to be subjected to sound source separation in each band, and the sound source-separated signal is inputted to a subband combining part 55 to calculate an output signal corresponding to the original signal. <P>COPYRIGHT: (C)2003,JPO

Description

DETAILED DESCRIPTION OF THE INVENTION [0001] [0001] The present invention relates to a signal extraction method and a signal extraction method.
Signal extraction device, signal extraction program and its program
For recording media on which the
Cannot be observed directly, and noise and other signals may overlap.
In a situation where it is observed in a folded state,
Technology for estimating the desired original signal, such as a speech recognition device
Input microphone and the speaker are separated into microphones
In situations where sound other than the target speaker's voice is picked up
However, the target speaker's voice is extracted and speech recognition with a high recognition rate is performed.
Signal extraction method and signal extraction device that can be applied,
Signal extraction program and recording medium storing the program
About the body. [0002] 2. Description of the Prior Art Knowledge of original signal and mixing process is used at all
The question of estimating multiple linearly mixed signals without
The title is Blind Source Separation:
BSS), but the invention of this application
Belonging to sound source separation technology. Statistical independence between signals
The technique of separating linearly mixed signals based on
Independent Component Analysis (ICA)
Is done. Like a sound pickup in a real sound field, the signal
Linearly mixed signal with impulse response convolved
The issue is [0003] (Equation 1) [0004] It is expressed as follows. here, xj: Signal observed by sensor j si: Signal of signal source i hji: P-tap input from signal source i to microphone j
Pulse response (linear system) It is. When blind source separation is practical,
Loose response length P is often large. For example, in a real environment
Reverberation is convoluted with speech
Will be. General conference room of 150-300ms
The impulse response length P is thousands of
Length. In independent component analysis, N signal sources
Signals emitted are assumed to be statistically independent of each other
And the observed signal obtained in the form of equation (1) and the length is Q taps
Separation filter group wijExtraction using a separation system consisting of
Put out. This separation filter group w ijSeparation and extraction using
Signal y obtained byi(N) [0006] (Equation 2) Is expressed as Figure 1 shows the above mixing and separation process.
Is a diagram for explaining a case where N = M = 2. FIG. Separation system
Is a learning rule wij k + 1= Wij k+ @ Wij k              (3) Output usingiAre estimated to be independent of each other
You. k indicates the number of learning updates. The problem is convolutional mixing
Is a complex problem, so the separation filter group can be found directly.
It is difficult to turn. Therefore, the discrete Fourier transform (DF
T) is often used to convert to the frequency domain.
You. This is called frequency domain blind sound source separation (frequency domain
Area BBS). First, the frequency of equation (1) is calculated by DFT.
Convert to area. X (ω, m) = H (ω) S (ω, m) (4) This allows the convolutional mixing problem to be solved instantaneously at each frequency.
Can be expressed as a mixed problem, simplifying the problem
You. As described above, the estimation of the separation process is performed at each frequency.
Output signal Y1(Ω, m), YTwo(Ω, m) are independent of each other
Thus, if N = M = 2, a (2 × 2) separation matrix
It suffices to estimate W (ω). Y (ω, m) = W (ω) X (ω, m) (5) Reverberation time with a large impulse response length P as in a real environment is 1
In the case of about 50 to 300 ms or more, the length is about the same
It is necessary to find a separation filter having Therefore,
For finding the separation matrix W (ω) in the wavenumber domain BSS
Use a frame T longer than the impulse response length P of the room.
Need to perform DFT analysis and increase the number of frequency bins
There is. However, the learning data of the determined length
When analyzed using frames, the data at each frequency
Data at each frequency.
The properties deteriorate. This will be described with reference to FIG. H
In the case of FIG. 2A where the frame length is short,
Data number is sufficient, statistical properties at each frequency
Is fully guaranteed. But what can be estimated
The length of the separation filter is short and insufficient. on the other hand,
In the case of FIG. 2B having a long frame length, a long separation filter is used.
Can be prepared, but the number of data at each frequency
The statistical nature of the data deteriorates because of the small number.
The problem that the statistical properties deteriorate will be described with reference to FIG.
I do. FIG. 3 shows all correlation coefficients γω (ω is a subscript).
Averaged at the frequency of [0011] [Equation 3] FIG. When the frame length T is large
In addition, the correlation between the signals increases,
It can be seen that the assumption of standing is broken. Thus, the frequency
The number domain BSS has a length longer than the impulse response length P of the room.
Perform DFT analysis using frame T to increase the number of frequency bins
Need to be added, but doing so at each frequency
Separation becomes difficult because the statistical properties of the data deteriorate. [0013] SUMMARY OF THE INVENTION Conventional frequency domain BS
In S, a large frame length T corresponding to a long reverberation
The correlation between signals is high when using
Independence assumption required for the application is broken and high performance cannot be obtained
Was. Therefore, the present invention particularly has a large impulse response length P.
In a BSS with thousands of thresholds, multiple mixed signals are supported.
Output signal corresponding to the original signal
To maintain the assumption of signal independence in each band.
Signal for estimating a separation filter long enough for separation.
Signal extraction method, signal extraction device, signal extraction program
It provides a recording medium on which the program is recorded.
You. [0014] The present invention has a long impulse response.
Signal from multiple mixed signals observed through the path
Signal extraction and separation methods based on their independence
Input a plurality of mixed signals to the sub-band analysis unit.
Subband analysis is performed on each of N (N: integer) bands, and
Time of the band corresponding to the signal of each band subjected to the band analysis
Input to the area BSS section to separate sound source for each band,
Input the separated signal to the sub-band synthesizer to support the original signal
A signal extraction method for obtaining an output signal to be performed is configured. And a path having a long impulse response.
The original signal from multiple mixed signals observed through
In a signal extraction device that separates and extracts based on
N mixed signals (N: integer)
A sub-band analysis unit for analyzing the
Time domain BSS for sound source separation for each analyzed signal in each band
The unit is equipped with a sound source separated signal and supports the original signal.
Signal having a sub-band synthesizer for obtaining an output signal to be output
An extraction device was configured. Here, each of the mixed signals is
Sub-band analysis into N (N: integer) bands
The signal of each band subjected to the command analysis is separated into sound sources for each band.
Output signal corresponding to the original signal from the signal separated from the sound source in the band
Configure a signal extraction program that performs
did. Then, a plurality of mixed signals are N each.
(N: Integer) sub-band analysis and sub-band analysis
The signal of each analyzed band is separated into sound sources for each band, and
Find the output signal corresponding to the original signal from the signal separated from the sound source
Record a signal extraction program that
A recording medium was constructed. [0017] DETAILED DESCRIPTION OF THE INVENTION The present invention utilizes sub-band analysis.
And perform signal separation in each band. This is the sub-band BS
Called S. Subband BSS is the number of subbands to be divided
Since the number can be freely selected, it is shown in FIG.
Bandwidth that satisfies statistical properties in each subband
The number of divisions can be selected. And the frequency domain BS
S can estimate only one tap filter at each frequency
Although not possible, the sub-band BSS is shown in FIG.
So that each band can have a long filter
From the point of view, I saw it in full band even if the number of divisions was small
Sometimes a sufficiently long filter can be estimated. With the above two points, the sub-band BSS is used.
Therefore, even if the impulse response length P is long,
Responding to reverberation while preserving the statistical properties of the data
Can be estimated. [0019] FIG. 4 shows an embodiment of the present invention.
Will be explained. FIG. 4 is a diagram showing the entire sub-band BSS.
It is. (1) Subband analysis process First, the input observation signal x1(n), xTwo(n) is the sub
Band analyzer 511, 51TwoEnter the sub-van
Is analyzed. (2) Sound source separation process Then, the observation signal x1(n), xTwoSubbands for each band of (n)
The component of the signal analyzed by the command
Time domain BSS unit 53 of each corresponding band1...
・, 53NAnd the sound is separated. (3) Subband synthesis process Finally, the components of the signal separated into each band are divided into each time domain B
SS unit 531, ..., 53N From the sub-band synthesis unit 55
1, 55TwoInput to the signal s1(n), sTwoThe signal corresponding to (n)
No. y1, YTwoAre synthesized and output. As mentioned above,
Multiple mixtures observed via paths with impulse response
Separate and extract the original signal from the combined signal based on its independence
The signal extraction device according to the present invention inputs a plurality of mixed signals.
To analyze each of N (N: integer) bands
And a signal of each band subjected to sub-band analysis.
A time domain BSS unit 53 for sound source separation for each sound source,
Input signal to obtain an output signal corresponding to the original signal.
And a band combining unit 55. Next, a detailed description will be given with reference to FIG.
First, the subband analysis process will be described. (1)
The sub-band analysis process is performed by the sub-band analyzer 51.1And 5
1TwoAnd the SSB modulator 521And 52TwoIs composed of
ing. Assuming that the number of band divisions is N and the thinning rate is M, x
jThe signal after the decimation in the k-th band of (n) is [0022] (Equation 4) Is calculated. Where WN= Exp (j
2π / N). H (n) is the band used for analysis.
Low-pass filter in the range [-π / N, π / N],     h (n) = {sin (n / N)} / (n / N) (8) Is used. At this time, X (k, m) is obtained as a complex number.
However, the time domain BSS unit 53 in the sound source separation process at the subsequent stage
1, ..., 53NIs a real number algorithm,
In order to handle signals in real numbers, for example, SSB (single side wave)
A band) modulator 52 can be used. SSB modulator
The subband using 52 is the alias in the frequency domain.
When the number of band divisions is N in order to avoid
= N / 4. A real number signal by SSB modulation is X
j ssbWhen expressed as (k, m),   Xj ssb(K, m) = Re [Xj(K, m)] cos (mπ / 2)                     + Im [Xj(K, m)] sin (mπ / 2) (9) Is obtained by Next, the sound source separation process will be described.
You. (2) The sound source separation process includes N time domain BSS units 53
1, ..., 53NIt consists of. Here, it is used for each band.
As an example of the time domain BSS algorithm used, the signal
Derived from the evaluation function based on the nonstationarity of
You. For simplicity, the output signal Yi ssb(K, n) is y
iAbbreviated as (n). If the cross-correlation of the output signal is
Non-negative evaluation function that takes the minimum value 0 when it becomes 0 in [0026] (Equation 5) Consider the following. Here, y (n) = [y
1(n), yTwo(n)]T Is the output signal and Ry b(Τ)
Is the covariance matrix of the output signal (= <y (n) yT(n-
τ)>b ) And <x>bIs a block b (b = 1,...)
.., B) are time averages. Separation filter w
ijIs obtained by differentiating this evaluation function Q with w (k).
In the calculated natural gradient, further improve the performance
R fory bNot only (0), but also time-lag correlation
Ry bBy taking into account (τ), it can be obtained as follows. [0028] (Equation 6)Equation (11) is derived by T. Nishikawa, H. Saruw
atari, K. Shikano, “Comparison ofblind source separ
ation methods based on time-domain ICA using nonst
ationarity and multistage ICA, ”IEICE Tech.Rep., Ja
n.2002. With this update formula
Wij(Element of w) and using equation (2)
To obtain a separated signal. This separated signal is represented by Y in FIG.
i ssb(K, n). The above method uses quadratic statistics.
Method using higher-order statistics is adopted.
It is also possible to use. Finally, the subband synthesis process will be described.
I do. (3) The subband synthesis process is performed by the SSB demodulation unit 5
41, 54TwoAnd the sub-band combining section 551, 55TwoAnd by
Be composed. First, the signal Y separated in each bandi ssb(K,
m) is converted to the SSB demodulation unit 541, 54TwoDemodulate by SSB
You. Re [Yi(K, m)] = Yi ssb(K, m) cos (m
π / 2) Im [Yi(K, m)] = Yi ssb(K, m) sin (m
π / 2) This demodulated signal Yi(K, m) is the sub-band synthesis unit 5
Input to 5, [0031] (Equation 7) The signal y synthesized byi(n) is obtained
You. Here, f (n) is the band [-π /
M, π / M].     f (n) = {sin (n / M)} / (n / M) (13) Is used. Here, an operation flowchart of the embodiment will be described.
As shown in FIG. And the above signal extraction device
May be configured with an electronic computer as a main component.
Also, the present invention can be downloaded from CDs or other storage media.
Downloaded or downloaded via communication line
Installing and running the ram on this computer
Can be. [0033] As described above, according to the present invention,
If each band has a long
Since the filter can be estimated, the high
Release performance is expected. FIG. 6 shows this sub-band BSS
The effect is shown. Here, the number of band divisions N =
64, decimation rate M = 16, length Q of separation filter of each band
su b = 64. This is the frequency domain for the thinning rate M
This is equivalent to BSS “32”, and the filter length is full bar.
Equivalent to a 1024 tap separation filter. Experiment
Is the case where the reverberation of the room is 150 ms and 300 ms
I followed. In FIG. 6, the numbers on the horizontal axis represent the frequency domain B.
The frame length (= filter length) in the SS
Results from frequency domain BSS using frame length
You. “SUB” is the result of sub-band BSS.
is there. In the frequency domain BSS, a long separation of 1024 lengths
The performance was degraded when seeking the filter.
Determining the length of 1024 separation filters in BSS
And high separation performance is obtained. Note that the original signal
The value of the correlation coefficient of equation (6) for the sub-band division is
0.028 for male and male voices, 0.018 for male and female voices, female voice
It was 0.020 in a female voice. Therefore, the independence assumption is
You can think that it is kept enough.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a diagram showing a signal separation model. FIG. 2 is a view for explaining the statistical properties and filter length of each frequency band. FIG. 3 is a diagram showing that the assumption of independence is broken at a large frame size in the frequency domain BSS. FIG. 4 is a view for explaining an embodiment of a sub-band BSS. FIG. 5 is a diagram for explaining details of an embodiment of a subband BSS. FIG. 6 is a view for explaining effects of the embodiment. 7 is a flowchart of an embodiment. [Description of References] 51 Sub-band analysis unit 53 Time-domain BSS unit 55 Sub-band synthesis unit

   ────────────────────────────────────────────────── ─── Continuation of front page    (72) Inventor Ryo Mukai             2-3-1 Otemachi, Chiyoda-ku, Tokyo Sun             Within the Telegraph and Telephone Corporation (72) Inventor Hiroshi Sawada             2-3-1 Otemachi, Chiyoda-ku, Tokyo Sun             Within the Telegraph and Telephone Corporation (72) Inventor Hiroshi Saruwatari             8916-5 Takayama, Ikoma City, Nara Prefecture D-307 F-term (reference) 5D015 EE05

Claims (1)

  1. Claims: 1. A signal extraction method for separating and extracting an original signal from a plurality of mixed signals observed via a path having a long impulse response on the basis of its independence. The signal is input to the sub-band analysis unit, and is subjected to sub-band analysis into N (N: integer) bands, and the signals of the respective sub-bands are subjected to time domain B of the corresponding band
    A signal extraction method comprising: inputting a signal to an SS unit to separate a sound source for each band; and inputting the separated signal to a subband synthesis unit to obtain an output signal corresponding to an original signal. 2. A signal extracting apparatus for separating and extracting an original signal from a plurality of mixed signals observed via a path having a long impulse response based on its independence, wherein a plurality of mixed signals are inputted and N (N: integer) sub-band analysis unit, and a time-domain BSS unit for sound source separation for each sub-band-analyzed signal. A signal extraction device comprising a subband synthesis unit for obtaining an output signal corresponding to a signal. 3. A sub-band analysis of each of a plurality of mixed signals into N (N: integer) bands, a signal of each band subjected to the sub-band analysis is subjected to sound source separation for each band, and a signal obtained by sound source separation to each band. A signal extraction program for obtaining an output signal corresponding to the original signal from 4. A signal obtained by subjecting a plurality of mixed signals to sub-band analysis into N (N: integer) bands, separating the signals of each band subjected to the sub-band analysis for each band, and separating the sound source for each band. A recording medium on which is recorded a signal extraction program for obtaining an output signal corresponding to the original signal from the program.
JP2002072111A 2002-03-15 2002-03-15 Method, device and program for extracting signal, and recording medium recorded with the program Pending JP2003271168A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2002072111A JP2003271168A (en) 2002-03-15 2002-03-15 Method, device and program for extracting signal, and recording medium recorded with the program

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP2002072111A JP2003271168A (en) 2002-03-15 2002-03-15 Method, device and program for extracting signal, and recording medium recorded with the program

Publications (1)

Publication Number Publication Date
JP2003271168A true JP2003271168A (en) 2003-09-25

Family

ID=29202189

Family Applications (1)

Application Number Title Priority Date Filing Date
JP2002072111A Pending JP2003271168A (en) 2002-03-15 2002-03-15 Method, device and program for extracting signal, and recording medium recorded with the program

Country Status (1)

Country Link
JP (1) JP2003271168A (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2006030754A1 (en) * 2004-09-17 2006-03-23 Matsushita Electric Industrial Co., Ltd. Audio encoding device, decoding device, method, and program
KR100600313B1 (en) 2004-02-26 2006-07-14 남승현 Multipath is a method and an apparatus for the separation of a frequency domain blind channel mixed signal
KR100612870B1 (en) 2004-11-10 2006-08-14 삼성전자주식회사 Appratus and method for seperating implusive event
JP2007226036A (en) * 2006-02-24 2007-09-06 Nippon Telegr & Teleph Corp <Ntt> Signal separation device, signal separation method, signal separation program, and recording medium, and signal direction-of-arrival estimation device, signal direction-of-arrival estimation method, signal direction-of-arrival estimation program, and recording medium
EP1895515A1 (en) * 2006-07-28 2008-03-05 Kabushiki Kaisha Kobe Seiko Sho Sound source separation apparatus and sound source separation method
US8095493B2 (en) 2007-01-31 2012-01-10 Sony Corporation Information processing apparatus, information processing method and computer program

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100600313B1 (en) 2004-02-26 2006-07-14 남승현 Multipath is a method and an apparatus for the separation of a frequency domain blind channel mixed signal
JP2007526511A (en) * 2004-02-26 2007-09-13 ナム,スン ヒョンNAM Seung Hyon Method and apparatus for blind separation of multipath multichannel mixed signals in the frequency domain
WO2006030754A1 (en) * 2004-09-17 2006-03-23 Matsushita Electric Industrial Co., Ltd. Audio encoding device, decoding device, method, and program
JP4809234B2 (en) * 2004-09-17 2011-11-09 パナソニック株式会社 Audio encoding apparatus, decoding apparatus, method, and program
US7860721B2 (en) 2004-09-17 2010-12-28 Panasonic Corporation Audio encoding device, decoding device, and method capable of flexibly adjusting the optimal trade-off between a code rate and sound quality
KR100612870B1 (en) 2004-11-10 2006-08-14 삼성전자주식회사 Appratus and method for seperating implusive event
JP4630203B2 (en) * 2006-02-24 2011-02-09 日本電信電話株式会社 Signal separation device, signal separation method, signal separation program and recording medium, signal arrival direction estimation device, signal arrival direction estimation method, signal arrival direction estimation program and recording medium
JP2007226036A (en) * 2006-02-24 2007-09-06 Nippon Telegr & Teleph Corp <Ntt> Signal separation device, signal separation method, signal separation program, and recording medium, and signal direction-of-arrival estimation device, signal direction-of-arrival estimation method, signal direction-of-arrival estimation program, and recording medium
EP1895515A1 (en) * 2006-07-28 2008-03-05 Kabushiki Kaisha Kobe Seiko Sho Sound source separation apparatus and sound source separation method
US7650279B2 (en) 2006-07-28 2010-01-19 Kabushiki Kaisha Kobe Seiko Sho Sound source separation apparatus and sound source separation method
US8095493B2 (en) 2007-01-31 2012-01-10 Sony Corporation Information processing apparatus, information processing method and computer program

Similar Documents

Publication Publication Date Title
JP5209033B2 (en) Partial complex modulation filter bank
Araki et al. The fundamental limitation of frequency domain blind source separation for convolutive mixtures of speech
US7797153B2 (en) Speech signal separation apparatus and method
US8150065B2 (en) System and method for processing an audio signal
JP4210521B2 (en) Noise reduction method and apparatus
JP2013527493A (en) Robust noise suppression with multiple microphones
KR20060046450A (en) Gain-constrained noise suppression
Schobben et al. A frequency domain blind signal separation method based on decorrelation
US8781137B1 (en) Wind noise detection and suppression
CN1460323A (en) Sub-and exponential smoothing noise canceling system
WO2007100330A1 (en) Systems and methods for blind source signal separation
CN1307747A (en) Convolutive blind source separation using multiple decorrelation method
US7158933B2 (en) Multi-channel speech enhancement system and method based on psychoacoustic masking effects
US8107631B2 (en) Correlation-based method for ambience extraction from two-channel audio signals
US20090086998A1 (en) Method and apparatus for identifying sound sources from mixed sound signal
Kim et al. Independent vector analysis: definition and algorithms
KR20060048954A (en) Method and apparatus for multi-sensory speech enhancement
US7711553B2 (en) Methods and apparatus for blind separation of multichannel convolutive mixtures in the frequency domain
DE60025748T2 (en) Voice recognition
US8467538B2 (en) Dereverberation apparatus, dereverberation method, dereverberation program, and recording medium
Shamsunder et al. Multichannel blind signal separation and reconstruction
US8724798B2 (en) System and method for acoustic echo cancellation using spectral decomposition
Park et al. Subband-based blind signal separation for noisy speech recognition
Smaragdis Discovering auditory objects through non-negativity constraints
KR20120063514A (en) A method and an apparatus for processing an audio signal

Legal Events

Date Code Title Description
A621 Written request for application examination

Free format text: JAPANESE INTERMEDIATE CODE: A621

Effective date: 20040210

A131 Notification of reasons for refusal

Effective date: 20050614

Free format text: JAPANESE INTERMEDIATE CODE: A131

RD03 Notification of appointment of power of attorney

Effective date: 20050812

Free format text: JAPANESE INTERMEDIATE CODE: A7423

A521 Written amendment

Free format text: JAPANESE INTERMEDIATE CODE: A523

Effective date: 20050812

A02 Decision of refusal

Free format text: JAPANESE INTERMEDIATE CODE: A02

Effective date: 20060620