ptrack.1

speech signal process tools

.\" @(#)ptrack.1	1.2	5/12/86
.TH PTRACK 1\-SPS "5/12/86"
.ds ]W "\fI\s+4\ze\h'0.05'e\s-4\v'-0.4m'\fP\(*p\v'0.4m'\ Entropic Processing, Inc.
.SH NAME
ptrack \- Performs pulse-to-pulse pitch tracking 
.SH SYNOPSIS
.B ptrack
[
.BI \-x " debug_level"
] [
.BI \-p " param_fil"
] [
.BI \-h " history_fil"
]
.I sdfile pitchfile
.SH DESCRIPTION
.PP
.I Ptrack
takes as input an SPS sampled data file and produces as output a pitch file.
See
[1]
for a description of the pitch tracker.
An SPS pitch file contains records holding the pitch locations, the length
of the pitch pulse, and the distance corresponding to the point
of maximum coherence, which is called the raw pulse distance.
.SH SPS PARAMETERS
The following parameters are read from the parameter file:
.IP
.sp
.I "start - integer"
.IP
This is the starting point in the sampled data file to begin pulse tracking:
note that an initial buffer of points are skipped beyond this point.
.sp
.I "nan - integer"
.IP
This is the number of points in the sampled data file to process.
.sp
.I "stepsz - integer"
.IP
Step size during unvoiced speech.
.sp
.I "wflt_order - integer"
.IP
The order of the filter used in whitening the data.
.sp
.I "p_method - string"
.IP
Only one is defined at this time: "C_PDM".
.sp
.I "u_pflt_order - integer"
.IP
Only zero is supported at this time.
.sp
.I "v_pflt_order - integer"
.IP
Only zero is supported at this time.
.sp
.I "pt_wsize - integer"
.IP
Size of the pulse tracking window.
.sp
.I "uv_thresh - float"
.IP
The pulse coherence threshold; it is used to make a raw voiced/unvoiced decission.
.sp
.I "min_pp - integer"
.IP
The minimum pulse duration allowed.
.sp
.I "min_coh_len - integer"
.IP
The minimum data segment length used for coherence estimation.
Currently this is equivalent to 
.I min_pp.
.sp
.I "max_pp - integer"
.IP
The maximum pulse duration allowed.
.sp
.I "max_coh_len - integer"
.IP
The maximum data segment length used for coherence estimation.
This option is not implemented.
.sp
.I "coh_method - string"
.IP
The type of coherence estimation method to use.
This option is not yet supported;
currently autocorrelations are used.
.sp
.I "lpf_num - vector"
.IP
A three element vector showing the numerator of a pole-zero filter - 
for example, {1.0, 2.0, 3.0}.
This option is not yet supported.
.sp
.I "lpf_den - vector"
.IP
A three element vector showing the denominator of a pole-zero filter - 
for example, {1.0, 2.0, 3.0}.
This option is not yet supported.
.sp
.I "win_type - string"
.IP
The type of window to use in the data analysis;
allowable values are "HAMMING" and "NONE".
This option is not yet supported.
/* NOTE: This option not relevant for ptrack---ssn*/
.sp
.I "dcrem - string"
"YES", "yes", "Y", and "y" removes the average value from the data before
pulse tracking.;
"NO", "N", "no", "n" leaves the data alone.
This option is not yet supported.
.sp
.I "coh_data - string"
.IP
This parameter specifies the type of data to be used for coherence computation.
"RAW" specifies raw data;
"RESIDUAL" specifies whitened speech (or residual signal);
"BOTH" specifies that coherence estimation is done for both residual and raw speech.
For best performance, use "BOTH". DSP pulse tracker is in some 
sense comparable to
pulse tracker working with "RESIDUAL" option. 
.sp
.I "coh_window_frac - float"
.IP
Fraction of the data used in computing the coherence.
This option is not yet supported.
.sp
.I "pb_offset - integer"
.IP
The number of 
points before the estimated pulse location that the coherence estimation
window is placed.
This option is not yet supported.
.sp
.I "harmonic_mult - float"
.IP
The weighting factor applied to the coherence estimate at local
maxima.
This option is not yet supported.
.sp
.I "tm_uv_thresh - float"
.IP
Coherence threshold to use for pulse tracking during voiced segments.
This option is not yet supported.
.sp
.I "am_uv_thresh - float"
.IP
Coherence threshold to use for pulse tracking during unvoiced to voiced transition.
This option is not yet supported.
.sp
.I "hp_thresh - float"
.IP
A threshold for high periodicity detection.
This option is not yet supported.
.sp
.I "gain_halt - float"
.IP
If the residual power falls below this value,
no further whitening of this data segment is done.
This option is not yet supported.
.sp
.I "pulse_srch - integer"
.IP
Range around the maximum coherence distance that is searched for pulses.
This option is not yet supported.
.sp
.PP
See [1] and PIT(5-SPS) for a description of the parameters.
.SH SPS HEADER
A new file header is created for the 
.I pit
file.
It contains values for all the parameters mentioned in 
.I SPS PARAMETERS.
In addition the new file header contains a value for each of the following
items:
.I src_sf, prog, vers, progdate, 
and
.I ndrec.
Finally,
the source sampled data file name is written in 
.I refer,
and the source file header is added as a source in the output file header.
.SH OPTIONS
The following options are supported:
.TP
.BI \-x " debug_level"
If 
.I debug_level
is positive,
.I ptrack
prints various information or debugging messages, on the standard error.
The messages are more verbose when
.B debug_level
is larger (up to 3).  If
.I debug_level
is 0, no message is printed.
The default is 0.
.TP
.BI \-p " param_fil"
uses the parameter file 
.I param_fil
rather than the default, which is
.I params.
.TP
.BI \-h " history_fil"
causes
.I ptrack
to write history information on the named file.
.br
.SH  FILES
.PP
.I ./params
is the default parameter file name.
.SH BUGS
If the input data is all zeros, program bombs!!
.SH REFERENCES
[1] ETM-S-85-29 
.I EPI Pulse Tracking Algorithm (PTRACK),
S. Shankar Narayan, 4/7/86
.SH FUTURE CHANGES
.PP
Efforts to improve the algorithm will continue.
Modifications will occur when we include silence detection in the
algorithm.
.SH SEE ALSO
dsptrack (1), ana (1), synt (1), PIT(5), SPS(5)