1SPHINX_CONT_SEG(1)          General Commands Manual         SPHINX_CONT_SEG(1)
2
3
4

NAME

6       sphinx_cont_seg - Segment a waveform file into non-silence regions
7

SYNOPSIS

9       sphinx_cont_seg [ options ]...
10

DESCRIPTION

12       This  program  reads an input file and segments it into individual non-
13       silence regions. It can process either file or read  data  from  micro‐
14       phone. Use following arguments:
15
16       -adcdev
17              of audio device to use for input.
18
19       -alpha Preemphasis parameter
20
21       -argfile
22              file giving extra arguments.
23
24       -dither
25              Add 1/2-bit noise
26
27       -doublebw
28              Use double bandwidth filters (same center freq)
29
30       -frate Frame rate
31
32       -infile
33              of audio file to use for input.
34
35       -input_endian
36              Endianness  of  input data, big or little, ignored if NIST or MS
37              Wav
38
39       -lifter
40              Length of sin-curve for liftering, or 0 for no liftering.
41
42       -logspec
43              Write out logspectral files instead of cepstra
44
45       -lowerf
46              Lower edge of filters
47
48       -ncep  Number of cep coefficients
49
50       -nfft  Size of FFT
51
52       -nfilt Number of filter banks
53
54       -remove_dc
55              Remove DC offset from each frame
56
57       -remove_noise
58              Remove noise with spectral subtraction in mel-energies
59
60       -remove_silence
61              Enables VAD, removes silence frames from processing
62
63       -round_filters
64              Round mel filter frequencies to DFT points
65
66       -samprate
67              Sampling rate
68
69       -seed  Seed for random number generator; if less than  zero,  pick  our
70              own
71
72       -singlefile
73              a single cleaned file.
74
75       -smoothspec
76              Write out cepstral-smoothed logspectral files
77
78       -transform
79              Which  type  of  transform  to use to calculate cepstra (legacy,
80              dct, or htk)
81
82       -unit_area
83              Normalize mel filters to unit area
84
85       -upperf
86              Upper edge of filters
87
88       -vad_postspeech
89              Num of silence frames to keep after from speech to silence.
90
91       -vad_prespeech
92              Num of speech frames to keep before silence to speech.
93
94       -vad_startspeech
95              Num of speech frames to trigger vad from silence to speech.
96
97       -vad_threshold
98              Threshold for decision between noise and  silence  frames.  Log-
99              ratio between signal level and noise level.
100
101       -verbose
102              Show input filenames
103
104       -warp_params
105              defining the warping function
106
107       -warp_type
108              Warping function type (or shape)
109
110       -wlen  Hamming window length
111

AUTHOR

113       Written  by  M.  K.  Ravishankar <rkm@cs.cmu.edu>.  This (rather lousy)
114       manual page by David Huggins-Daines <dhuggins@cs.cmu.edu>
115

117       Copyright © 1999-2001 Carnegie Mellon University.  See the file COPYING
118       included with this package for more information.
119
120
121
122                                  2008-05-12                SPHINX_CONT_SEG(1)
Impressum