1mlib_SignalMelCepstral_S16_mAeddpi(a3LMiLbIBL)ibrarymlFiubn_cStiigonnaslMelCepstral_S16_Adp(3MLIB)
2
3
4
6 mlib_SignalMelCepstral_S16_Adp - perform cepstral analysis in mel fre‐
7 quency scale
8
10 cc [ flag... ] file... -lmlib [ library... ]
11 #include <mlib.h>
12
13 mlib_status mlib_SignalMelCepstral_S16_Adp(mlib_s16 *cepst,
14 mlib_s32 *cscale, const mlib_s16 *signal, void *state);
15
16
18 The mlib_SignalMelCepstral_S16_Adp() function performs cepstral analy‐
19 sis in mel frequency scale. The scaling factor of the output data will
20 be calculated based on the actual data.
21
22
23 The first two steps of mel scale cepstral analysis is the same as in
24 general cepstral anaysis. After the logarithm of the spectrum magnitude
25 is obtained, it is converted into mel frequency scale before the
26 inverse Fourier transform.
27
28 +-----------+ +-----------+
29 | Linear | | Inverse |
30 ... ------>| to |------->| Fourier |----->
31 X'(k) | Mel Scale | X''(m) | Transform | c(n)
32 +-----------+ +-----------+
33
34
35
36 where X'(k) is defined in linear frequency scale and X''(m) is defined
37 in mel frequency scale.
38
39
40 The mel frequency scale is defined as following.
41
42 freq_mel = melmul * LOG10(1 + freq_linear / meldiv)
43
44
45
46 where freq_mel is the frequency in mel scale, freq_linear is the fre‐
47 quency in linear scale, melmul is the multiplying factor, muldiv is the
48 dividing factor.
49
50
51 Optionally, a bank of band pass filters in linear frequency scale can
52 be used below the bank of band pass filters in mel frequency scale, as
53 shown below in linear frequency scale.
54
55 0 f1 f2 f3 fp fp+1 fp+2 fp+3 fp+q
56 |---|---|---| ... |---|----|-----| ... | ... -> freq
57
58
59
60 where fp = melbgn, fp+q = melend, p = nlinear, q = nmel; the filters
61 number 1 to p are defined in linear frequency scale which have equal
62 bandwidth in linear frequency scale; the filters number p+1 to p+q are
63 defined in mel frequency scale which have equal bandwidth in mel fre‐
64 quency scale and increasing bandwidth in linear frequency scale.
65
66
67 See Digital Signal Processing by Alan V. Oppenheim and Ronald W.
68 Schafer, Prentice Hall, 1974.
69
70
71 See Fundamentals of Speech Recognition by Lawrence Rabiner and Biing-
72 Hwang Juang, Prentice Hall, 1993.
73
75 The function takes the following arguments:
76
77 cepst The cepstral coefficients.
78
79
80 cscale The scaling factor of cepstral coefficients, where
81 actual_data = output_data * 2**(-scaling_factor).
82
83
84 signal The input signal vector, the signal samples are in Q15 for‐
85 mat.
86
87
88 state Pointer to the internal state structure.
89
90
92 The function returns MLIB_SUCCESS if successful. Otherwise it returns
93 MLIB_FAILURE.
94
96 See attributes(5) for descriptions of the following attributes:
97
98
99
100
101 ┌─────────────────────────────┬─────────────────────────────┐
102 │ ATTRIBUTE TYPE │ ATTRIBUTE VALUE │
103 ├─────────────────────────────┼─────────────────────────────┤
104 │Interface Stability │Committed │
105 ├─────────────────────────────┼─────────────────────────────┤
106 │MT-Level │MT-Safe │
107 └─────────────────────────────┴─────────────────────────────┘
108
110 mlib_SignalMelCepstralInit_S16(3MLIB), mlib_SignalMelCep‐
111 stral_S16(3MLIB), mlib_SignalMelCepstralFree_S16(3MLIB), attributes(5)
112
113
114
115SunOS 5.11 2 Mar 20m0l7ib_SignalMelCepstral_S16_Adp(3MLIB)