.TH ENVANAL 1carl CARL
.SH NAME
envanal \- data reduction by hierarchical syntactic function analysis.
.SH SYNOPSIS
.B envanal
[
.B flags
] <
.B floatsams
>
.B output
.SH DESCRIPTION
Flags: (default values in parenthesis)
.TP
.B -dN
set time-domain length greater than or equal to which
an input segment causes a break (.01).
.TP
.B -yN
set amplitude difference greater than
which an input segment will cause a break (.01).
.TP
.B -sN
set maximum size to which a segment can grow before causing a break (.1).
.TP
.B -DN
pass 2 version of -d (.1).
.TP
.B -YN
pass 2 version of -y (.2).
.TP
.B -SN
pass 2 version of -s (.4).
.TP
.B -nN
max. # of segments expected (8K segments).
.TP
.B -1
output pass one analysis only.
.TP
.B -2
output both passes (default).
.TP
.B -v
verbose mode.
.TP
.B -g
produce gen1 mode output, text [x,y] pairs.
.TP
.B -f
produce floatsam output (output must be file or pipe).
.TP
.B -pN
produce crude CRT plot of output, skipping by N (1 sample).
.TP
.B -PN
enable peak detector with threshold of N (.01).
.TP
.B -RN
set sampling rate to N.
.TP
.B -h
print help message.
.PP
This program is based on Strawn's algorithm [1] of syntactic analysis.
.PP
It implements a two level grammar for a data reduction by
the recursive recognition of line segment features of
an input waveform.  It produces on the standard output either a 
summary of these features, a plot of the resulting data-reduced function,
or the floating point binary representation of that function.
In addition, a peak-reading algorithm has been added that causes the
analysis to proceed along the tops of the input segments.
.SH AUTHOR
Gareth Loy
.SH SEE ALSO
.IP [1]
John Strawn, ``Approximation and Syntactic Analysis of Amplitude
and Frequency Functions for Digital Sound'', CMJ, V4, #3.
.SH BUGS
The number of expected segments must be supplied.  
If the input data exceeds this, the program bombs.
Since this number is one
of the things the analysis hopes to yield, this is a tautology.  Meanwhile,
one is stuck with guessing.
