man.diff

来自「speech signal process tools」· DIFF 代码 · 共 318 行

.\" Copyright (c) 1987, 1989 Entropic Speech, Inc. All rights reserved.
.\" @(#)auto.1	1.9	4/2/97 ESI
.TH AUTO 1\-ESPS 4/2/97
.ds ]W "\fI\s+4\ze\h'0.05'e\s-4\v'-0.4m'\fP\(*p\v'0.4m'\ Entropic Speech, Inc.
.SH "NAME"
auto \- Compute autocorrelation sequence of real data
.SH "SYNOPSIS"
.B auto
[
.BI \-l " frame_len"
] [
.BI \-o " order"
] [
.BR \- { pr } "\fI range\fP"
] [
.BI \-w " window_type"
] [
.BI \-x " debug_level"
] [
.BI \-P " params"
] [
.BI \-S " step_size"
] [
.B  \-B
]
.I sd.in
.I fea_ana.out
.SH "DESCRIPTION"
.PP
.I auto
takes as input a single-channel, real ESPS sampled-data (FEA_SD or SD) file, 
.I sd.in,
and computes the positive-lag elements of the power-normalized
autocorrelation function of one or more fixed-length segments to
produce an ESPS FEA_ANA file
.I fea_ana.out
containing one or more records.  (The autocorrelation function for a
real signal is even, so the negative lag elements are equal to the
positive lag elements. That is, R[1] = R[\-1], R[2] = R[\-2], etc.)
.PP
All input frames have the same length (see \fB\-l\fR below).
The initial points of successive frames are separated by the step size
specified by \fB\-S\fR. Thus the three cases \fIstep < frame_len,
step = frame_len\fR, and \fIstep > frame_len\fR,
correspond to overlapping frames, exactly abutted frames,
and frames separated by gaps (sometimes called underlapping frames).
.PP
In addition to the normalized autocorrelation values, each record
contains the power corresponding to the input data segment, the length
of the data segment, and the starting point for that record in the
input sampled data file.  By default the power computed for each
record is the power corresponding to the unwindowed data.  This can be
changed to the power of the windowed data by using the \fIpower\fR
parameter in the params file. See \fIpower\fR under \s-1\fBESPS
PARAMETERS\fR\s+1 for more details.
.PP
By default, a lag-product form is used to compute the autocorrelation.
If the \fB-B\fP is used, a structured covariance method is used.  
This is usually provides a much better estimate, especially for 
short, noisy data sequences.  
.PP
If the input file name
.I sd.in
is replaced by "\-", then \fIstdin\fR is read; similarly, if
.I fea_ana.out
is replaced by "\-", then the output is written to \fIstdout\fR.
.SH "OPTIONS"
.PP
The following options are supported:
.TP
.BI \-l " frame_len" " \fR[0]\fP"
Specifies the length of each frame, and overrides the 
.I frame_len
value that may be in the parameter file.
The number of frames processed
is the largest number with the given length
.RI ( frame_len )
and spacing
(see
.BR \-S )
that will fit within the range
(determined by \fB\-p\fP option or
.I start
and
.I nan
in the parameter file or from ESPS Common).
If 
.I frame_len
is zero, 
.I auto
sets the frame length
equal to the total range \- i.e., a single frame is processed.
This holds whether 
.I frame_len
comes from the \fB\-l\fP option or from the parameter file.  
It also holds if the \fB\-l\fP option is not used and the
.I frame_len
parameter is not given in the parameter file.  
.TP
.BI \-o " order" " \fR[10]\fP"
Specifies the order of the autocorrelation function.  Autocorrelation
lags one through \fIorder\fR are stored for each record in the
\fIspec_param\fR field of the FEA_ANA record.  If the number of data
points in each frame (\fIframe_len\fR) is less than the \fIorder\fR,
\fIauto\fR warns and exits.  The default order is 10.
.TP
.BI \-p " first" : "last"
.TP
.BI \-p " first" :+ "incr" 
In the first form, a pair of unsigned integers specifies the range of
sampled data to analyze.  If 
.IR last " = " first " + " incr,
the second form (with the plus sign) specifies the same range as the
first.  If 
.I first
is omitted, the default value of 1 is used.  Both forms of the option
override the values of
.I start
and
.I nan
in the parameter file or ESPS Common file.  If the \fB\-p\fP option is not
used, the range is determined from the ESPS Common file if the appropriate
parameters are present (see ESPS COMMON).  
.TP
.BI \-r " first" : "last"
.TP
.BI \-r " first" :+ "incr" 
The
.B \-r
option is synonymous with
.BR \-p .
.TP
.BI \-w " window_type" " \fR[RECT]\fP"
Specifies the data window to apply to the data.  The default window
type is a rectangular window with amplitude equal to one.  Possible
window types include the following: rectangular("RECT"), Hamming
("HAMMING"), Hanning ("HANNING"), cosine to the fourth power ("COS4"),
and triangular ("TRIANG").  See the window(\3-ESPSsp) manual page for
a complete list of window functions.
.TP
.BI \-x " debug_level" " \fR[0]\fP"
Specifies that debugging output be printed on stderr; larger values
of 
.I debug_level
result in more output.  The default is for no output.  
.TP
.BI \-P " param" " \fR[params]\fP"
Specifies the name of the parameter file. The default name is
\fIparams\fR.
.TP
.BI \-S " step_size" " \fR[frame_len]\fP"
Initial points of consecutive frames differ by this number of samples.
If this option is omitted, the parameter file is consulted, and if no
value is found in the parameter file, a default equal to
\fIframe_len\fR is used.  This results in exactly abutted frames.
.TP
.B \-B 
Specifies that the "best" autocorrelation method be used -- i.e., a
structured covariance method, rather than the standard lag-product
algorithm.
.SH ESPS PARAMETERS
.PP
The parameter file is not required to be present, as there are 
default parameter values that apply.  If the parameter file 
does exist, the following parameters are read:
.IP
.I "start \- integer"
.IP
The first point in the input sampled-data file that is processed.  A
value of 1 denotes the first sample in the file.  This is only read
if the \fB\-p\fP option is not used.  If it is not in the parameter
(or Common) file, the default value of 1 is used.  
.sp
.I "nan \- integer"
.IP
The total number of points to analyze.  If no value is given, the
value is set equal to the total number of points in the input file
minus \fIstart\fR + 1.  Both the parameter file value and the default
value may be overridden by means of the \fB\-p\fP option.
.sp
.I "frame_len \- integer"
.IP
The number of points analyzed for each frame.  If no value is given,
or the value is zero,
the value is set equal to the range (i.e., so that only a single frame is
processed).  This value is not read if the command line option \fB\-l\fP is
used.  Both the parameter file value and the default value may be
overridden by means of the \fB\-l\fP option.  To see how the number of
frames is determined see the \fB\-l\fP option.
.sp
.I "step_size \- integer"
.IP
Initial points of consecutive frames differ by this number of samples.
This parameter is not read if the \fB\-S\fR option is specified.
If the option is omitted and no value is found in the parameter
file,
a default equal to \fIframe_len\fR is used (resulting
in exactly abutted frames).
.sp
.I "order \- integer"
.IP
The order of the autocorrelation function.
If no value is given in the file, a default value of 10 is used.
This value is not read if the command line option
\fB\-o\fP is used.
.sp
.I "window \- string"
.IP
Specify a data window to apply to the sampled data before computing the
autocorrelation
function.
This value is not read if the command line option \fB\-w\fP
is used.
If the option is omitted and no value is found in the parameter file,
the default used is a rectangular window with amplitude one.
.sp
.I "power \- string"
.IP
Specify the method for computing the power in each record.
Two methods are supported:
\fIunwindowed\fR (the default) and \fIwindowed\fR.
For \fIunwindowed\fR,
the power corresponding to the input data before any windowing
is computed;
for \fIwindowed\fR,
the power corresponding to the windowed data is computed.
.sp
.I "strcov_auto \- string"
.IP
Specify whether ("yes") or not ("no") to use a structured covariance
method rather than a lag-product method.  This parameter is not 
read if the \fB-B\fP option is used.  The default is not to use 
structured covariance.  
.PP
The values of parameters obtained from the parameter file are printed
if the environment variable ESPS_VERBOSE is 3 or greater.  The default
value is 3.
.SH "ESPS COMMON"
.PP
It the \fB\-p\fP option is not used and ESPS Common is enabled, the
parameters
.I start
(starting point) and 
.I nan
(number of points) are read from ESPS Common if they exist and 
if the Common parameter
.I filename 
matches
.I sd.in. 
If Common processing is enabled and if
.I fea_ana.out
is not standard output, the Common parameters 
.I "filename, prog, start, step_size, frmlen,"
and
.I nan
are written to Common, where 
.I filename
is set to the input 
.I sd.in.
.PP
ESPS Common processing may be disabled by setting the environment variable
USE_ESPS_COMMON to "off".  The default ESPS Common file is .espscom 
in the user's home directory.  This may be overridden by setting
the environment variable ESPSCOM to the desired path.  User feedback of
Common processing is determined by the environment variable ESPS_VERBOSE,
with 0 causing no feedback and increasing levels causing increasingly
detailed feedback.  If ESPS_VERBOSE is not defined, a default value of 3 is
assumed.
.SH ESPS HEADERS
.PP
.I auto
reads the value of 
.I common.type
from the input sampled-data file 
.I sd.in.  
.PP
Relevant fields in the type-dependent portion of the 
output file header are filled appropriately.  
.PP
Two generic header items that describe the processing 
are written to the output file header:
\fIdata_window \- coded\fR and \fIstep_size \- integer\fR.
\fIdata_window\fR
contains the name of the type of data window used on the data
prior to the computation of the autocorrelation function;
\fIstep_size\fR contains the number of points between consecutive frames.
In addition,
the generic header item \fIstart_time\fP (type DOUBLE) is written in
the output file.  The value written is computed by taking the
\fIstart_time\fP value from the header of the input file (or zero, if
such a header item doesn't exist) and adding to it the offset time
(from the beginning of the input file) of the first point processed
plus one half of \fIframe_len\fP.  (Thus, \fIstart_time\fP is in the middle of
the first frame, which is appropriate since the output data represent
the entire frame; without this adjustment for \fIframe_len\fP,
\fIwaves\fP+ displays would not line up properly.) 
.SH "FUTURE CHANGES"
.PP
Add more data window types.
.SH "SEE ALSO"
.PP
.nf
refcof(1\-ESPS), cross_cor(1\-ESPS), me_spec(1\-ESPS), rem_dc(1\-ESPS), 
transpec(1\-ESPS), spectrans(1\-ESPS), toep_solv(1\-ESPS), window(3\-ESPS), 
FEA_SD(5\-ESPS), SD(5\-ESPS), FEA_ANA(5\-ESPS), ESPS(5\-ESPS), 
\fIget_auto\fP (3\-\s-1ESPSsp\s+1), \fIstrcov_auto\fP (3\-\s-1ESPSsp\s+1)
.fi
.SH "BUGS"
.PP
None known.
.SH "AUTHOR"
.PP
Manual page and program by Dave Burton.
Conversion from SD to FEA_SD by Rod Johnson.  
Structured covariance added by Shore.