Vous êtes sur la page 1sur 14

Linear predictive coding of

speech (LPCS)
Presented By Leena Jena
1
st
Year M.Tech (Branch:Signal
Processing)
29-10-2014 Leena Jena
What is LPCS?

Linear predictive coding (LPC) is Tool.

Speech processing is the study of speech signals and the
processing methods of these signals. The signals are usually
processed in a digital representation, so speech processing
can be regarded as a special case of digital signal processing.

29-10-2014 Leena Jena
Why LPCS?
LPC is one of the most powerful speech analysis
techniques, and one of the most useful methods
for encoding good quality speech at a low bit rate
and provides extremely accurate estimates of
speech parameters.
29-10-2014 Leena Jena
Block Diagram of simplified model of
speech production
29-10-2014 Leena Jena
LPC Analysis Equations
. 1
) (
) (
) (
) ( ) ( ) ( ) ( ) (
). ( ) (
). ( ) ( ) (
1
1
~
1
~
1

=
=
=
= =
= =
=
+ =
p
k
k
k
p
k
k
p
k
k
p
k
k
z a
z S
z E
z A
k n s a n s n s n s n e
k n s a n s
n Gu k n s a n s
29-10-2014 Leena Jena
LPC Analysis Equations
. ) ( ) (
) (
) ( ) (
) ( ) (
2
1
2

=
=
+ =
+ =
= m
p
k
n k n n
m
n n
n
n
k m s a m s E
m e E
m n e m e
m n s m S

We seek to minimize the mean squared error signal:

29-10-2014 Leena Jena
p i k i a i
k m S i m S k i
k m S i m S a m s i m s
p k
a
E
p
k
n
k
n
m
n n n
m m
n n
p
k
k n n
k
n
,..., 2 , 1 ) , ( ) 0 , (
) ( ) ( ) , (
) ( ) ( ) ( ) (
,..., 2 , 1 , 0
1
1
= =
=
=
= =
c
c


=
.
=
.
| |
|
Terms of short-term covariance:
( * )
With this notation, we can write (*) as:
A set of P equations, P unknowns
29-10-2014 Leena Jena


=
.
=
.
.
=
=
p
k
n
k
n
m
n n
p
k
k
m
n n
k a
k m s m s a m
s E
1
1
2
). , 0 ( ) 0 , 0 (
) ( ) ( ) (
| |
The minimum mean-squared error can be expressed as:
LPC Analysis Equations
29-10-2014 Leena Jena
The Autocorrelation Method
.
0
1
), ( ) ( ) , (
0
1
), ( ) ( ) , (
) (
. , 0
1 0 ), ( ). (
) (
) ( 1
0
1
0
1
0
2
p k
p i
k i m s m s k i
p k
p i
k m s i m s k i
m e E
otherwise
N m m w n m s
m s
k i N
m
n n n
p N
m
n n n
p N
m
n n
n
s s
s s
+ =
s s
s s
=
=

s s +
=


=
+
=
+
=
|
|
The mean squared error is
w(m): a window zero outside 0mN-1
And:
29-10-2014 Leena Jena
The Autocorrelation Method
) ( ) , (
: function ation autocorrel simple to reduces function
covariance the k, - i of function a only is ) , ( Since
.
0
1
), ( ) ( ) , (
) ( 1
0
k i r k i
k i
p k
p i
k i m s m s k i
n n
n
k i n
m
n n n
=
s s
s s
+ =


=
|
|
|
29-10-2014 Leena Jena
The Autocorrelation Method
.
) (
) 3 (
) 2 (
) 1 (
) 0 ( . . . ) 3 ( ) 2 ( ) 1 (
) 3 ( ... ) 0 ( ) 1 ( ) 2 (
) 2 ( ... ) 1 ( ) 0 ( ) 1 (
) 1 ( ... ) 2 ( ) 1 ( ) 0 (
: as form matrix in expressed be can and
1 ), ( |) (|
: ) ( ) ( i.e.
symmetric, is function ation autocorrel the Since
2
1
1
(
(
(
(

=
(
(
(
(
(

(
(
(
(

s s =
=
.
.
.
=
.

p r
r
r
r
a
a
a
r p r p r p r
p r r r r
p r r r r
p r r r r
p i i r a k i r
so k r k r
n
n
n
n
p
n n n n
n n n n
n n n n
n n n n
p
k
n
k
n
n n
29-10-2014 Leena Jena
The Covariance Method

=
s s
s s
+ =
s s
s s
=
=
s s
1
1
0
1
0
2
.
0
1
), ( ) ( ) , (
, variables of change by or,
0
1
), ( ) ( ) , (
: as defined ) , ( with
) (
: directly speech unweighted the use to
and 1 0 error to computing of interval the change
i N
i m
n n n
n
N
m
n n
n
N
m
n n
p k
p i
k i m s m s k i
p k
p i
k m s i m s k i
k i
m e E
N m
|
|
|
29-10-2014 Leena Jena
The Covariance Method
.
) 0 , (
) 0 , 3 (
) 0 , 2 (
) 0 , 1 (
) , ( ) 3 , ( ) 2 , ( ) 1 , (
) , 3 ( ) 3 , 3 ( ) 2 , 3 ( ) 1 , 3 (
) , 2 ( ) 3 , 2 ( ) 2 , 2 ( ) 1 , 2 (
) , 1 ( ) 3 , 1 ( ) 2 , 1 ( ) 1 , 1 (
3
2
1
(
(
(
(
(
(

=
(
(
(
(
(
(
(

(
(
(
(
(
(

.
.
.
.
p
a
a
a
a
p p p p p
p
p
p
n
n
n
n
p
n n n n
n n n n
n n n n
n n n n
|
|
|
|
| | | |
| | | |
| | | |
| | | |

The resulting covariance matrix is symmetric, but not Toeplitz,and can be solved efficiently by
a set of techniques called Cholesky decomposition
29-10-2014 Leena Jena
Application of LPC
Speech coding/synthesis
Pitch prediction (Used in Codebook Excited
Linear Prediction)
Computer Music:cross synthesis
Spectral Envelope Estimation

29-10-2014 Leena Jena

Vous aimerez peut-être aussi