iir_filt.1

来自「speech signal process tools」· 1 代码 · 共 253 行

.\" Copyright (c) 1986-1990 Entropic Speech, Inc.
.\" Copyright (c) 1991 Entropic Research Laboratory, Inc.; All rights reserved
.\" @(#)iir_filt.1	1.13 27 Sep 1997 ESI/ERL
.TH IIR_FILT 1\-ESPS 27 Sep 1997
.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
iir_filt \- design recursive filter.
.SH SYNOPSIS
.B iir_filt
[
.BI \-P " param_file"
]
[
.BI \-x " debug_level"
]
.I filt_file
.SH DESCRIPTION
.PP
.I Iir_filt
designs a recursive filter (often called an infinite impulse response 
filter)
and puts the coefficients, poles, and zeros in the
.IR FEA_FILT (5\-ESPS)
file
.I filt_file.  
It supports filter reponse types of lowpass, highpass,
bandpass, and bandstop filters.  The design methods supported are
Butterworth, Chebyshev1, Chebyshev2, and elliptical methods.  Other input
parameters are passband and stopband frequency edges, and the frequency 
versus loss characteristics.
.PP
All the design information must be supplied in the parameter
file. By using \fIeparam\fR(1-ESPS), however,
the user will be prompted for all the design information. 
.PP
If
.I filt_file
is "\-", the output goes to standard output.
.SH OPTIONS
The following options are supported:
.TP
.BI \-P " param_file"
The file 
.I param_file
is used for the parameter file instead of the default, which is
.I params.
.br
.TP
.BI \-x " debug_level"
A value of 0 (the default value) will cause
.I iir_filt
to work silently, unless there is an error.
A nonzero value will cause various parameters to be printed out
during the filter design. A value greater than 9 will cause all the
debug messages to be printed.
.br
.SH ESPS PARAMETERS
.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.
.PP
.IP
.I "samp_freq \- float"
.IP
The value of the sampling frequency.
.IP
.I "gain \- float"
.IP
The gain of the filter in the pass band.
.IP
.I "filt_method \- string"
.IP
The type of the polynomial used in the filter design.
The options are BUTTERWORTH, CHEBYSHEV1, CHEBYSHEV2, and ELLIPTICAL.
.IP
.I "filt_type \- string"
.IP
The type of filter response to design. Allowable types are the following:
low pass (LP), high pass (HP), band pass (BP), and band 
stop (BS).
.IP
.I "pass_band_loss \- float"
.IP
Specifies the maximum dB deviation from 1 for the filter amplitude response
within the pass band.  
.IP
.I "stop_band_loss \- float"
.IP
Specifies the minimum dB loss for the filter amplitude response in the 
stop band from the pass band.
.IP
.I "p_freq1 \- float"
.IP
Also referred to as "passband frequency 1".  For lowpass and highpass filters, 
this is the passband frequency. For bandpass
and bandstop filters, this is the first passband frequency.  See \fBEXAMPLES\fR below.
.IP
.I "s_freq1 \- float"
.IP
Also referred to as "stopband frequency 1".  Stop band frequency for low 
pass and high pass filters.  Or the first
stop band frequency for band pass and band stop filters.  See \fBEXAMPLES\fR below.
.IP
.I "p_freq2 \- float"
Also referred to as "passband frequency 2".  The second pass band 
frequency for band pass and band stop filters.  This is
ignore for lowpass and highpass filters.  See \fBEXAMPLES\fR below.
.IP
.I "s_freq2 \- float"
Also referred to as "stopband frequency 2".  The second stop band 
frequency for band pass and band stop filters.  This is
ignore for lowpass and highpass filters.  See \fBEXAMPLES\fR below.
.IP
.I "filt_order \- int"
.IP
This is an optional parameter.  Only use this parameter if the optimal 
filter order computed by the program to meet the filter specification is
not desired.  If \fIfilt_order\fR is present, \fIiir_filt\fR uses this
parameter for its filter order.  A warning message is printed, values
for the optimal filter order and \fIfilt_order\fR are printed.
.PP
.SH ESPS COMMON
Esps Common is not processed.
.sp .5
.SH ESPS HEADER
A new FEA_FILT 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 
FEA_FILT type 
.IP
.I max_num
.IP
It is set equal to \fIfilt_order\fR.
.IP
.I max_denom
.IP
It is set equal to \fIfilt_order\fR.
.IP
.I func_spec
.IP
It is set equal to IIR.
.IP
.I type
.IP
It is set equal to the specified \fIfilt_type\fR value.
.IP
.I method
.IP
It is set equal to the specified \fIfilt_method\fR value.
.IP
.I define_pz
.IP
It is  set to YES.
.IP
.I pass_band_loss
.IP
It is set equal to \fIpass_band_loss\fR.
.IP
.I stop_band_loss
.IP
It is set equal to \fIstop_band_loss\fR.
.IP
.I filt_order
.IP
It is the filter order.
.PP
.SH WARNING
The stopband loss of the Elliptical filter may not meet
the desired specification in some cases.  This is due to the 
approximation procedure 
involved in the Elliptical filter design method that maximizes
the stopband attenuation, but gives the desired passband attenuation, 
passband frequencies, and stopband frequencies.  In these cases,
raise the stopband loss, \fIstop_band_loss\fR.
.PP
The filter response can be plotted and viewed using \fIfiltspec(1-ESPS)\fR and
\fIplotspec(1-ESPS)\fR, for example,
.PP
\fIfiltspec filt_file - | plotspec - \fR
.PP
.SH FUTURE CHANGES
A generic header item \fIdelay_samples\fP, which has the
"filter delay" in samples, will be added. This will enable programs
to automatically compensate for the delay caused by the filter and allow
plotting programs to automatically time align waveforms. 
.PP
For now, \fIimpulse_resp\fP(1\-\s-1ESPS\s+1) and \fIplotsd\fP(1\-\s-1ESPS\s+1) 
can be used to compute and plot the impulse response, which can be visually
inspected to locate the peak of the response. The offset of the peak from
the beginning of the response is often useful as an estimate of the delay.
\fIaddgen\fP(1\-\s-1ESPS\s+1) can then be used to add the \fIdelay_samples\fP
generic header item.
.PP
.SH EXAMPLES
The parameters \fIp_freq1, s_freq1, p_freq2\fR, and \fIs_freq2\fR are 
used in the following manner for each filter response type.  Suppose the
Nyquist rate is denoted as \fInf\fR.  
.PP
.nf
              passband (Hz)      stopband (Hz)
 Lowpass      0 to p_freq1      s_freq1 to nf   
 Hihgpass     0 to s_freq1      p_freq1 to nf   

              stopband1          passband              stopband2
 Bandpass     0 to s_freq1      p_freq1 to p_freq2     s_freq2 to nf

              passband1          stopband              passband2
 Bandstop     0 to p_freq1      s_freq1 to s_freq2     p_freq2 to nf   

.ni
.PP
The easiest way to use this program is by \fIeparam(1-ESPS)\fR.  Depending
on the filter reponse type (e.g. lowpass) selected, \fIeparam(1-ESPS)\fR
prompts for appropriate pass/stopband frequency edges (e.g. p_freq1) 
from left to right.  Simply enter frequencies in an ascending order as 
they are prompted.  For examples, suppose a bandpass filter,
.nf

   %eparam iir_filt bandpass.filt
   Desired filter type: [ELLIPTICAL]  (Choices are: BUTTERWORTH CHEBYSHEV1 
   CHEBYSHEV2 ELLIPTICAL):
   Desired filter reponse [LP]  (Choices are: LP HP BP BS):BP
   Desired sampling frequency. [8000] :
   Stopband frequency 1 [1100] :1000
   Passband frequency 1 [1000] :1100
   Passband frequency 2 [2100] :2000
   Stopband frequency 2 [2000] :2100
   Desired maximum pass band loss (dB). [1] :
   Desired minimum stop band attenuation (dB). [20] :

.fi.
This creates a bandpass filter with passband over 1100 to 2000 Hz, and
stopband over 0 to 1000 Hz and 2100 to 4000 Hz.
.PP
.SH SEE ALSO
.nf
\fIESPS\fP(5\-ESPS), \fIFEA\fP(5\-ESPS), \fIFEA_FILT\fP(5\-ESPS), \fIfilter2\fP(1\-ESPS), 
\fIwmse_filt\fP(1\-ESPS), \fInotch_filt\fP(1\-ESPS), \fIxpz\fP(1\-ESPS)
.fi
.SH REFERENCES
.PP
Parks, T. W. and Burrus, C. S.,
\fIDigital Filter Design\fR, 
John Wiley and Sons, 1987, New York
.PP
Sublett, B. and Burton, D.,
\fIESPS APPLICATIONS NOTE: Filtering Sampled Data\fR,
ESPS USER'S MANUAL
.SH AUTHOR
Derek Lin