1mlib_SignalLPCCovariance_S1m6e(d3iMaLLIiBb)Library Fmulnicbt_iSoingsnalLPCCovariance_S16(3MLIB)
2
3
4

NAME

6       mlib_SignalLPCCovariance_S16,  mlib_SignalLPCCovariance_S16_Adp  - per‐
7       form linear predictive coding with covariance method
8

SYNOPSIS

10       cc [ flag... ] file... -lmlib [ library... ]
11       #include <mlib.h>
12
13       mlib_status mlib_SignalLPCCovariance_S16(mlib_s16 *coeff,
14            mlib_s32 cscale, const mlib_s16 *signal, void *state);
15
16
17       mlib_status mlib_SignalLPCCovariance_S16_Adp(mlib_s16 *coeff,
18            mlib_s32 *cscale, const mlib_s16 *signal, void *state);
19
20

DESCRIPTION

22       Each function performs linear predictive coding with covariance method.
23
24
25       In linear predictive coding (LPC) model, each speech sample  is  repre‐
26       sented as a linear combination of the past M samples.
27
28                      M
29              s(n) = SUM a(i) * s(n-i) + G * u(n)
30                     i=1
31
32
33
34       where  s(*)  is the speech signal, u(*) is the excitation signal, and G
35       is the gain constants, M is the order of the linear prediction  filter.
36       Given  s(*),  the  goal is to find a set of coefficient a(*) that mini‐
37       mizes the prediction error e(*).
38
39                             M
40              e(n) = s(n) - SUM a(i) * s(n-i)
41                            i=1
42
43
44
45       In covariance method, the coefficients can be obtained by solving  fol‐
46       lowing set of linear equations.
47
48               M
49              SUM a(i) * c(i,k) = c(0,k), k=1,...,M
50              i=1
51
52
53
54       where
55
56                       N-k-1
57              c(i,k) =  SUM s(j) * s(j+k-i)
58                        j=0
59
60
61
62       are  the  covariance coefficients of s(*), N is the length of the input
63       speech vector.
64
65
66       Note that the covariance matrix R is a symmetric matrix, and the  equa‐
67       tions can be solved efficiently with Cholesky decomposition method.
68
69
70       See  Fundamentals  of Speech Recognition by Lawrence Rabiner and Biing-
71       Hwang Juang, Prentice Hall, 1993.
72
73
74       Note for functions with adaptive scaling (with _Adp postfix), the scal‐
75       ing  factor  of  the output data will be calculated based on the actual
76       data; for functions with non-adaptive scaling (without  _Adp  postfix),
77       the  user  supplied  scaling factor will be used and the output will be
78       saturated if necessary.
79

PARAMETERS

81       Each function takes the following arguments:
82
83       coeff     The linear prediction coefficients.
84
85
86       cscale    The scaling factor of  the  linear  prediction  coefficients,
87                 where actual_data = output_data * 2**(-scaling_factor).
88
89
90       signal    The input signal vector with samples in Q15 format.
91
92
93       state     Pointer to the internal state structure.
94
95

RETURN VALUES

97       Each  function returns MLIB_SUCCESS if successful. Otherwise it returns
98       MLIB_FAILURE.
99

ATTRIBUTES

101       See attributes(5) for descriptions of the following attributes:
102
103
104
105
106       ┌─────────────────────────────┬─────────────────────────────┐
107       │      ATTRIBUTE TYPE         │      ATTRIBUTE VALUE        │
108       ├─────────────────────────────┼─────────────────────────────┤
109       │Interface Stability          │Committed                    │
110       ├─────────────────────────────┼─────────────────────────────┤
111       │MT-Level                     │MT-Safe                      │
112       └─────────────────────────────┴─────────────────────────────┘
113

SEE ALSO

115       mlib_SignalLPCCovarianceInit_S16(3MLIB),      mlib_SignalLPCCovariance‐
116       Free_S16(3MLIB), attributes(5)
117
118
119
120SunOS 5.11                        2 Mar 2007mlib_SignalLPCCovariance_S16(3MLIB)
Impressum