cb_filt.1

来自「speech signal process tools」· 1 代码 · 共 214 行

.\" Copyright (c) 1992 Entropic Research Laboratory, Inc.; All rights reserved
.\" @(#)cb_filt.1	1.5 4/2/97 ERL
.ds ]W (c) 1992 Entropic Research Laboratory, Inc.
.TH  CB_FILT 1\-ESPS 4/2/97
.SH NAME
.nf
cb_filt \- design a constraint-based FIR filter using the 
Simplex algorithm
.fi
.SH SYNOPSIS
.B
pkmc_filt
.BI \-P " param_file"
[
.BI \-x " debug_level"
]
.I feafilt_file
.SH DESCRIPTION
.PP
The program \fIcb_filt\fR designs a linear phase finite impulse response
(FIR) filter with its specifications defined in the parameter file 
\fIparam_file\fR.  The filter coefficients are saved in the output file
\fIfeafilt_file\fR.  If \fIfeafilt_file\fR is replaced by "-", the 
standard output is written.
.PP
The filter is designed using the Simplex algorithm to solve for a 
constraint-based problem.  The constraints are expressed in terms of 
upper/lower limits and concavity properties on the response, 
the algorithm finds the shortest
filter length which allows these constraints to be met, and then find
a filter of that order which is farthest from the upper and lower
constraint boundaries in a mini-max sense.
.PP
The constraints have the following forms:  bandedges, upper bounds and lower
bounds on the bands, concavity, and huggness constraints.  They are
expressed in various combinations of parameters in \fIparam_file\fR.
.PP
This program is more general than \fIpkmc_filt(1-ESPS)\fR.  It can
design filter with nearly flat passband magnitude response by specifying
the concavity property on the band.
.PP
Also see the shell script \fIxfir_filt(1-ESPS)\fR that is a cover
script for this and other FIR filter design programs.
.PP
.SH OPTIONS
.PP
The following option is supported:
.TP
.BI \-x " debug_level \fR[0]\fP"
If \fIdebug_level\fR is positive, \fIcb_filt\fR prints debugging messages 
and other information on the standard error output.  The messages 
proliferate as the \fIdebug_level\fR increases.  If \fIdebug_level\fP 
is 0 (the default), no messages are printed.  
.SH ESPS PARAMETERS
The following parameters are read from \fIparam_file\fR.
.TP
.I filt_length_L - int
The largest filter length allowed.  It must be less or equal to 128
.TP
.I filt_length_S - int
The smallest filter length allowed.  It must be greater or equal to 1.
\fIcb_filt\fR will finds the shortest filter length that satisfies the
constraints.  If fixed filter length is desired, set \fIfilt_length_L\fR
equal to \fIfilt_length_S\fR.
.TP
.I ngrid - int
This is an optional parameter. It specifies the number of grid points 
The default value is 10 * (\fIfilt_length_L\fR /2 ) + 1.
.TP
.I samp_freq - float
The sampling frequency.
.TP
.I nspec - int
Number of specifications.  Each band needs at least two specifications,
one for lower bound constraint, one for upper bound constraint, and 
possibly one for concavity constraint.  For example, a lowpass filter may
have \fInspec\fR equals 5 -- 2 constraints for upper and lower bounds in
passband, 2 constraints for upper and lower bounds in stopband, and 1
concavity constraint for passband.
.TP
.I model - string
Specifies the symmetry property of the designed FIR impulse response.
Use the value \fIcosine\fR for even symmetry -- Type I or Type II FIR for
multiband filters.  Use the value \fIsine\fR for odd symmetry -- Type
III or Type IV FIR for differentiators
.PP
The following set of the parameters have the forms of \fIspec[i]_XXX\fR,
where \fIi\fR denotes the specification number, \fIXXX\fR denotes
the kind of specification.  For example, 
\fIspec2_edge1\fR is a parameter for the left edge of the second 
specification.  \fIi\fR starts from 1.  
.TP
.I spec[i]_type - string
Type of constraint on the \fIi\fRth specification.  The value \fI"limit"\fR
denotes upper/lower bound limit specification for the \fIi\fRth spec.
The value \fI"concave"\fR denotes up/down concavity specification for
the \fIi\fRth spec.  If \fIspec[i]_type\fR is \fI"concave"\fR, then
only \fIspec[i]_sense\fR, \fIspec[i]_edge1\fR, and \fIspec[i]_edge2\fR
need be defined.  Concavity constraint to the passband of the
filter gives nearly flat frequency response.  Otherwise, an upper/lower
limit constraint gives an equiripple frequency response.
.PP

.TP
.I spec[i]_sense - string
If \fIspec[i]_type\fR is \fI"limit"\fR, a value of "+" denotes an upper
bound spec; or "-" for a lower bound spec.  If \fIspec[i]_type\fR is
\fI"concave"\fR, "+" denotes concave up, "-" for concave down.
.TP
.I spec[i]_edge1 - float
The left bandedge for the \fIi\fRth spec.
.TP
.I spec[i]_edge1 - float
The right bandedge for the \fIi\fRth spec.
.TP
.I spec[i]_bound1 -  float
The bound on \fIspec[i]_edge1\fR.
.TP
.I spec[i]_bound2 -  float
The bound on \fIspec[i]_edge2\fR.  Values in between \fIspec[i]_bound1\fR
and \fIspec[i]_bound2\fR over the \fIi\fRth spec will be interpolated
according to the parameter \fIspec[i]_interp\fR.
.TP
.I spec[i]_hug - string
If it is \fI"not hugged"\fR, the response on the \fIi\fRth specification 
will be pushed as far as possible from the specified bounds.  If this
constraint needs not be optimized, then use the value \fI"hugged"\fR.
.TP
.I spec[i]_interp - string
If it is \fI"arithmetic"\fR, frequency response values from 
\fIspec[i]_bound1\fR to \fIspec[i]_bound2\fR over the band 
\fIspec[i]_edge1\fR to \fIspec[i]_edge2\fR is interpolated arithmetically
(linearly); if it is \fI"geometric"\fR, values are interpolated
geometrically (linearly in decibels).
.TP
.I push_direction - string
If it is \fI"left"\fR, a set of bandedges are pushed as far left as
possible while still respecting the constraints for the fixed 
length (i.e. \fIfilt_length_L = filt_length_S\fR) specified by user.
If it is \fI"right"\fR, the set of bandedges are pushed right.  If
no pushing is desired, specify \fIneither\fR.
If \fIpush_direction\fR is set to \fI"right"\fR or \fI"left"\fR, the
parameters \fInspec_pushed\fR, \fIspec[j]_pushed\fR must be specified.
.TP
.I nspec_pushed - int
Number of bands to be pushed.
.TP
.I spec[j]_pushed - int
A single number corresponding the specification number that contains
the desired bandedges to be pushed.  For example, if \fIspec1_pushed\fR 
equals to 4, \fIspec2_pushed\fR equals 3 means the bandedges specified in
specification number 3 and 4 are to be pushed.  The specification
numbers can be assigned to \fIspec[j]_pushed\fR, \fI1 <= j <= nspec\fR,
in any arbitrary order.
.PP
.SH ESPS COMMON
No ESPS common parameter processing is used.
.PP
.SH ESPS HEADERS
A new FEAFILT header is created for the  output  file.   The
program  fills  in  appropriate values in the common part of
the header as well as the  following  generic  header  items
associated with the FEAFILT type.
.PP
The \fIcb_specs\fR generic header item is a string containing
\fInspec\fR specifications.  Each specification starts with \fISPEC[i]\fR,
followed by \fI"limit"/"concave"\fR, \fIsense\fR, left edge, right edge,
bound at left edge, bound at right edge, huggness, and interpolation mode.
.PP
The \fImodel\fR generic header item denotes whether \fIsine\fR or \fIcosine\fR
model is used.
.PP
If \fIpush_direction\fR parameter is \fI"left"\fR or \fI"right\fR,
the \fInspec\fR generic header item denotes numbers of specification pushed.
.PP
If \fIpush_direction\fR parameter is \fI"left"\fR or \fI"right\fR,
the \fIpush_direction\fR generic header item is a string that first 
starts with the band direction pushed, followed by specification numbers
that are pushed.
.PP
In addition, the generic header item \fIdelay_samples\fR  (type  DOUBLE) is
added  to the header. \fIDelay_samples\fR  is  equal  to  
(filter length  -  1)/2.   
This represents  the  delay  to  the  center  of  the peak of the
impulse response.

.PP
.SH FUTURE CHANGES
.PP
.SH EXAMPLES
See its parameter setting example in \fI$ESPS_BASE/lib/params/Pcb_filt\fR.
.PP
.SH ERRORS AND DIAGNOSTICS
.PP
.SH BUGS
.PP
None known.
.SH REFERENCES
"METEOR: A Constraint-Based FIR Filter Design Program", Kenneth Steiglitz,
Thomas Parks, and James Kaier.  IEEE Transaction on Signal Processing,
Vol. 40, No. 8, August 1992.
.PP
.SH "SEE ALSO"
.nf
xfir_filt(1\-ESPS), pkmc_filt(1\-ESPS), win_filt(1\-ESPS), 
notch_filt(1\-ESPS), FEA_FILT(5\-ESPS), atofilt(1\-ESPS), 
wmse_filt(1\-ESPS), iir_filt(1\-ESPS), sfconvert(1\-ESPS)
.nf
.PP
.SH AUTHOR
Thanks to Dr. Kenneth Steiglitz, Thomas Parks, and James Kaiser for
permission to use the original C codes.  Program is made ESPS-compatible
by Derek Lin.