1
HC VI 
KHOA QUỐC TẾ VÀ ĐÀO TẠO SAU ĐẠI HỌC
 TIỂU LUẬN
Môn: XỬ LÝ TÍN HIỆU NÂNG CAO
ĐỀ TÀI: BỘ LỌC THÍCH NGHI
Giáo viên hướng dẫn: TS. Nguyễn Ngọc Minh
Nhóm Học viên thực hiện: Lý Hoàng Sáng
Nguyễn Xuân Khánh
Nguyễn Ngọc Bá
Đào Văn Thái
Lớp cao học: M12CQDT02-B


Hà Nội, tháng 01 năm 2012
2

   hoi v 
th c s d u sai s
e[n]. Nu mong muu dc cc tiu
  n mt mch lc rt ni tich lc
c g
Hu hi mch lc Wiener. Trong
 t tng trng s cu sai s
c cc tin mt mch lc ti v liu

y, mch lc thit k hoc bc thc b
   t k  nh, cn ph   t s  ng
 d lich lc cn x  thit
k c mch lc Wiener cn phi bit th
sng h u d
i gian b
Mc ti  c thc hin
 c nhn thit cho vic thit k mch lc Wiener hoch
4

lc tu ng dng thc t u lc
s d li ra ca mch lu
ca mch lc theo mt s  dng kt qu c
 u ch s ca mch theo kiu lp. S d 
i gi  hiu cht thng
i vi th nghim lp, c s ca mch l
th t u ch t thi.
Trong phm vi tiu lu u mt s v n
 l
1. Bộ lọc thích nghi
1.1. Hệ thống FIR truyền thống

Ca mch l

a mch lc.
x[n] = [x
n
x
n-1
x
n-2
 x
n-N+1
]
T
ng s ca b l
w =

[w
0
w
1

N-1
]
T
i ra ca mch lc
0
[ ] w[ ] [ ]w [ ]
T
k
y n k x n k x n

2.1. Phương pháp LMS
B lu nhi
nhiu gii thu u qu cho b l i thut
LMS (Least Mean Square.). Gii thu s t lc. Phn
 gii thiu mt s gii thut thc hiu t ng
dng ca lt s ng h  gii quyt
 thi.
7

n MMSE (Minimum mean square error)
Gii thu  s ca mch lc FIR t
vic ti s u
  phc) v

xx
(m) = E[x(n)x
*
(n-m)]
u x(n) qua h thng cnh, tng
th lc FIR v s -1. Tu ra
ca mch l





.x(n-k)
Sai s 
e(n) = y(n)  -


-
1)x(n-k)]}}}
= 
2
y  2Re[




(l)]
yx
(l) +
 





(l)h(k) 
yx
(l-k)
Vi 
2
y = E[y(n)]
2
, 
yx
(l) = E[y(n).x*(n-a h s.
Ta thy rng sai s i vi
 s ca mch lc. Do vc tiu ca MSE hay J(hM) ta ln

lk
= 
xx
(l-
y

n t 
yz
-1. Gii tp h
 s ca b lc t
h
opt
= 
M-1
, 
y
t qu MSE cc tiu v s cho bi h
opt
= 
M-1
, 
y

J
min
= J(h
opt
) = 
2
[y 

(n) +


 
Vi h
M
 s mch lc ti ln lp th  c nhy
ln th ng cho ln lp th u h
M
c chn bt
ku d
n nh c tiu ca J(h
M
p d
ving dc nht ( Steepest - ng
dc nh
S(n) = -g(n). vi ln lp th 
g(n) = dJ(h
M
(n))/d h
M
(n)

M
. h
M
(n) - 
i mi ln l ca h
M
(n) theo

(n)
V      n t x(n-   -  y, vector
n 
g(n) = -2e(n).X*M(n)
Vi e(n) = y(n)  p hp M mu ti ln lp th 
i thut:
h
M
(n+1) = h
M
(n) + (n)e(n).X*M(n)
Gii thuc gi thut Stochastic-gradient- mt bi
 c nhy (n) c 
(n) c nh d thc hin cho c phn cng ln phn mm.
(n) c p vi vic thi theo th
i s thay  x
nhi thu
h
M
(n+1) = h
M
(n) + e(n).X*M(n)
Vi    c nhy c nh.
Gii thut LMS thc hic s dng rng
ng dnh mch li hn c
c nhiu.
2.2. Lọc thích nghi bằng biến đổi wavelet.
Bii wavelet ri rc b l
10


{
0,

1,

2,

3,

0
,
1
,
2
,
3
}
T
=C
1
.x
 :
3 2 1 0
3 2 1 0
3 2 1 0
1 0 3 2
1
3 2 1 0
3 2 1 0
3 2 1 0

2,

0

1

2

3
}
T

0,

1,

2,

3,

0

1

2

3
}
T







Kt hc biu y s 
y= C.x= C
2
.C
1
.x
D y rng C
1
c giao, C
2
u kin trc giao s 
2222
0 1 2 3
1cccc   

0 2 1 3
0c c c c

2222
0 1 2 3
1dddd   


.y C x

Vi :
1 1 0 ma x
. ;
QQ
C C C C C Q Q



Q
max
 smc   c   thuc   di       lc
max 2
1
(log )
1
M
Q floor
N




u kin trc giao t  :
2
2
1
( ); 0, ,
2

2
1
( ); 0, ,
2
N
n n k
nk
N
c d k k







2
2
1
( ); 0, ,
2
N
n k n
nk
N
c d k k





c
c
c
c









ng hp t
   
22
2 2 1
1
1; 0, ,
2
kk
N
c c k


  


t :
1 1 1 1

2 1 2 1 2
21
3 1 2 1 2
cos cos cos
cos sin sin
sin sin sin
sin cos cos
cc
cc
cc
cc
  
  
  
  


  


Vii dng ma trn :
(2)
2
3
(2) (1)
2
01
(2) (1)
2
10





14

 m rng cho 6 h s :
(3)
3
5
(3) (2)
3
43
(3) (2)
33
32
(3) (2)
33
21
(3) (2)
3
10
(3)
3
0
000
000
00
00
0 0 0













si n ; cos
j j j j
sk



 vi h s d, kt hc :
(3) (3) (3) (3) (3) (3)
5 4 3 2 1 0
(3) (3) (3) (3) (3) (3)
5 4 3 2 1 0
33
33
1 1 2 2
33


  



y tng bii ma trn wavelet th nhc biu dii dng :
1 1 2 3
. ( ). . ( ). . ( )C E R S R S R
  



j
 :
15

000000
000000
0 0 0 0 0 0
0 0 0 0 0 0
()
0 0 0 0 0 0
0 0 0 0 0 0
000000
000000
jj
jj
jj
jj
j



n d
0 1 0 0 0 0 0 0
0 0 1 0 0 0 0 0
0 0 0 1 0 0 0 0
0 0 0 0 1 0 0 0
0 0 0 0 0 1 0 0
0 0 0 0 0 0 1 0
0 0 0 0 0 0 0 1
00000000
S 













 lc cho h vi x t hp vi DSP biu s dng
wavelet, ng dc trong thc t.
3. Mô phỏng ứng dụng mạch lọc thích nghi
Tri AEC)
  ng dng ca b l dng trong vic tri
vang.
: Scott C. Douglas
Nội dung:
- Gii thiu
- 
- u gn
- u xa
- u Microphone
- B ln tn s (The Frequency-Domain Adaptive Filter  FDAF)
- ng tri li (Echo Return Loss Enhancement  ERLE)
- Nhng ng ca Nh 
-  ng tri li
Giới thiệu
Tring vi cuc hi thoi trong vi
phng cuc hi thon thiu microphone d(n) bao gm
u gn (near-u xa
17

(the far-end echoed speech signal dhat(n)). Mn loi b 
ng v vic tri dng b l
Đáp ứng xung trong phòng
         n t    
       g      s
dng b l n cht v ng xung vi
t l mu h th

title('Near-End Speech Signal');
set(gcf, 'Color', [1 1 1])
p8 = audioplayer(v,fs);
playblocking(p8); 19
Tín hiệu âm thanh đầu xa
Trong h th u gc truyn
u xa chuyn l
load farspeech
x = x(1:length(x));
dhat = filter(H,1,x);
plot(t,dhat);
axis([0 33.5 -1 1]);
xlabel('Time [sec]');
ylabel('Amplitude');
title('Far-End Echoed Speech Signal');
set(gcf, 'Color', [1 1 1])
p8 = audioplayer(dhat,fs);
playblocking(p8); 20 Tín hiệu Microphone thu được.


plot(t,v(n),'g');
axis([0 33.5 -1 1]);
ylabel('Amplitude');
title('Near-End Speech Signal');
subplot(3,1,2);
plot(t,d(n),'b');
axis([0 33.5 -1 1]);
ylabel('Amplitude');
title('Microphone Signal');
subplot(3,1,3);
plot(t,e(n),'r');
axis([0 33.5 -1 1]);
xlabel('Time [sec]');
ylabel('Amplitude');
title('Output of Acoustic Echo Canceller');
set(gcf, 'Color', [1 1 1])
p8 = audioplayer(e/max(abs(e)),fs);
playblocking(p8);
23 Tăng cường triệt tiêu âm quay trở lại (Echo Return Loss Enhancement – ERLE)
 n truy c u gu xa vang l
 u xa vang li, b d
mn (trong dB) tu. Ta th ng 30 dB ERLE
c hi t.
24