man.diff

来自「speech signal process tools」· DIFF 代码 · 共 458 行 · 第 1/2 页

This option forces all computation to be carried out in double precision
format. By default if this option is not specified, 
floating point operation is used for
input data types of FLOAT, LONG, SHORT, BYTE, FLOAT_CPLX, LONG_CPLX, 
SHORT_CPLX, and BYTE_CPLX; otherwise if input data type is DOUBLE or 
DOUBLE_CPLX,
double precision operation is used.
.TP
.B \-f
This option saves the lowpass filter coeffiecients in the header of output
file.
.TP
.BI \-x " debug_level \fR[0]\fP"
If 
.I debug_level
is positive,
.I sfconvert
prints debugging messages and other information on the standard error
output.  The messages proliferate as the  
.I debug_level
increases.  If \fIdebug_level\fP is 0 (the default), no messages are
printed.  
.SH ESPS PARAMETERS
The parameter file is not required to be present, as there are default
parameters that apply.  If the parameter file does exist, the following
parameters are read:
.TP
.I new_sample_freq - int
The sampling frequency of output file.  This parameter is not read if the
\fB-s\fR option is used.
.TP
.I deviation - float
The maximum tolerable deviation from \fInew_sample_freq\fR in percent.
This parameter is not read if the \fB-v\fR option is used.
.TP
.I start - integer
The first point in the input sampled-data file that is processed.  A value
of 1 denotes the first sample in the file.  If it is not in the parameter
(or Common) file, the default value of 1 is used.
.TP
.I nan - integer
The total number of points to analyze.  If it is set to 0 or if the value
is not given, \fInan\fR is set equal to the total number of points in the 
input file minus \fIstart\fR plus 1.  
.TP
.I corner_freq - float
The corner frequency of the lowpass filter designed for interpolation.
This parameter is not read if the \fB-c\fR option is used.  A value of zero
indicates the Nyquist rate of output file is used.
.TP
.I Kaiserflag - int
This paramter is optional.  
It is used to determine the window type used for the filter design
in case of ambiguity.  Set \fIKaiserflag\fR to 1 for the Kaiser window type,
or 0 for non-Kaiser window type.  If there is no \fB-w\fR on the command line,
\fIsfwin_type\fR is set to a non-Kaiser window type in the paramter file, 
and \fIKaiserflag\fR exists in the paramter file and it is 
equal to zero, then the window type determined by \fIsfwin_type\fR is
used for the filter design.  Otherwise, the Kaiser windowing filter 
design is used.  In the case when \fB-w\fR exists on the command line and
\fIKaiserflag\fR is 1, then window type specified by \fB-w\fR is used since
a command line option always takes precedence.
.I rej_db - float
The rejection ratio of the lowpass filter designed by Kaiser window 
from passband to stopband in dBs.  This paramter is ignored if a non-Kaiser
window type is used for filter design.
.TP
.I trans_band - float
The transition bandwidth of the lowpass filter designed by the Kaiser windowing
method between passband and stopband.
This paramter is ignored if a non-Kaiser
window type is used for filter design.
.TP
.I sfwin_type - string
Window type used in designing the lowpass filter other than Kaiser
windowing method.  Available parameters are WT_RECT (rectangular window), 
WT_HAMMING (Hamming window), WT_TRIANG (Triangular window),
WT_HANNING (Hanning window), and WT_COS4 (Cosine to power of 4 window).
.TP
.I sfwin_len - float
Window length in seconds used in designing the lowpass filter designed by
\fIsfwin_type\fR window type.
This paramter is ignored if the Kaiser windowing filter design method is used.
.TP
.I channels - string
This parameter specifies which channels from a multichannel input file for
sampling frequency conversion.  This parameter is not read if the \fB-e\fR
option is used.  
.TP
.I output_type - string
This parameter specifies the output data type.  Available data types are
DOUBLE, FLOAT, LONG, SHORT, BYTE, DOUBLE_CPLX, FLOAT_CPLX, LONG_CPLX, 
SHORT_CPLX, and BYTE_CPLX. This parameter is not read if the \fB-o\fR option
is used.
.TP
.I dflag - int
This specifies whether resampling is to be carried out in floating point or
double precision.  A value of \fI1\fR is for double precision.  
A value of \fI0\fR specifies floating point computation.
.PP
.SH ESPS COMMON
.PP
If the \fB-r\fR option is not used and ESPS Common processing is enabled,
the following items are read from the ESPS Common File provided that no 
input file is given on the command line and provided that standard input
is not used.
.IP
.I "filename - string"
.IP
This is the name of the input file.
.IP
.I "start - integer"
.IP
This is the starting point in the input file for frequency conversion.
.sp
.I "nan - integer"
.IP
This is the number of points in the input file for frequency conversion.
A value of zero means the last point in the file.
.sp
.PP
The following items are written into the ESPS Common file, provided
ESPS Common processing is enabled and the output file is not <stdout>.
.IP
.I "start - integer"
.IP
A value of 1 is written to the Common file.
.sp
.I "nan - integer"
.IP
The number of points in the output file.
.sp
.I "prog - string"
.IP
This is the name of the program (\fIsfconvert\fP in this case).
.sp
.I "filename - string"
.IP
The name of the output file.
.PP
.SH ESPS HEADERS
The generic header item \fIrecord_freq\fR  is written in the 
output file with the actual sampling frequency which may be different from
\fInew_sample_freq\fR if \fIdeviation\fR by the \fB-v\fR option is not zero. The
input file sampling frequency is also written as \fIsource_freq\fR.  Both
header items are type DOUBLE.  The corner frequency of lowpass filter is
written as \fIcorner_freq\fR as type DOUBLE if its default value 0 is not
used.

\fIstart_time\fR header item of type DOUBLE of output file is computed
by taking the \fIstart_time\fR of the input file and adding to it the 
offset time (from the beginning of the first input file) of the first
point used in frequency conversion.

The header item, \fIfilter_siz\fR, for the size of the lowpass filter is 
written as type SHORT.  If \fB-f\fR is specified on the command line,
the lowpass filter coeffiecients are saved in the header.  The coefficients, 
\fIfilter\fR, is 
written as type DOUBLE if input data is type DOUBLE or DOUBLE_CPLX,
if the \fB-d\fR option is set, or if the value of \fIdflag\fR in the 
parameter file is \fIyes\fR.  Otherwise \fIfilter\fR is type FLOAT.

If the Kaiser windowing method is used in filter design, \fIrej_db\fR and
\fItrans_band\fR are written as type FLOAT. 
Otherwise, \fIsfwin_type\fR and \fIsfwin_len\fR are written as type
CODED and FLOAT, respectively.

The header item, \fIsource_file\fR, is the input file name.  If the input
data is multichannel, channels selected for sampling frequency conversion
is written in the header item \fIchannels\fR.
.PP
.SH EXAMPLES
Converting the input sampling frequency 8000 Hz of input data channels
1, 3, 4, and 5 to 6000 Hz with the aliased components 80 dB down.

	\fIsfconvert -s6000 -R80 -e1,3:5 input.sd output.sd\fR

Dividing the input sampling frequency of 8000 Hz by one third (2666.66...).  
The conversion factor is 1/3.  However, using the option \fI-s 2666\fR for
new sampling frequency will result in large \fIup\fR and \fIdown\fR factors and
long filter length. The \fB-v\fR option 
allows \fIsfconvert\fR to find the exact intended conversion factor of 1/3
by allowing a small deviation from the necessarily truncated
\fInew_sample_freq\fR.  
The following command finds the actual sampling frequency to be 2666 
Hz within plus or minus 26.66 Hz, or exactly 8000/3 Hz with the values
of 1 and 3 for the \fIup\fR and \fIdown\fR ratios. 

	\fIsfcovnert -s2666 -v1 input.sd output.sd\fR

Converting the input sampling frequency 8000 Hz to 6000 Hz with a lowpass
filter of corner frequency of 2500 Hz instead of the default 3000 Hz, Nyquist
frequency.

	\fIsfcovnert -s6000 -c2500 input.sd output.sd\fR

Frequency conversion to 12000 Hz using a lowpass filter designed by Hamming 
window for input data sampled at 8000 Hz.  The filter will be 0.02 second
long, or 160 points in samples.

	\fIsfconvert -s12000 -wWT_HAMMING -l0.02 input.sd output.sd\fR

.PP
.SH ERRORS AND DIAGNOSTICS
In the case when the interpolation factor \fIup\fR is very large and window 
length \fIsfwin_len\fR supplied by \fB-l\fR is too short such that 
no convolution is possible,
an error message is issued and the program exits.
.PP
.SH BUGS
.PP
None known.
.SH REFERENCES
Alan V. Oppenheim and Ronald W. Schafer, \fIDiscrete-Time Signal Processing\fP.
Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 1989
.PP
.SH "SEE ALSO"
filter(1-ESPS), demux(1-ESPS), type_convert(3-ESPS), window(3-ESPS) 
.PP
.SH AUTHOR
.PP
Program and man page by Derek Lin