synt.1

speech signal process tools

.\"@(#)synt.1	1.5 1/22/87
.TH SYNT 1\-SPS 1/22/87
.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"
synt \- Performs pitch-synchronous synthesis
.SH "SYNOPSIS"
.B synt
[
.B \-d
] [
.BI \-x " debug_level"
] [
.BI \-h " his_file"
] [
.BI \-P " param_file"
] [
.BI \-s
]
.VI \-p " range"
.I ana_file sd_file
.SH DESCRIPTION
.PP
.I Synt
takes as input an SPS ANA file 
.I ana_file
containing analysis information, performs pitch-synchronous synthesis,
and outputs the synthesized speech to an SPS SD file 
.I sd_file.
If the
.B \-s
option is specified, the output file is written as shorts, otherwise
it is written in floating point format.
.SH SPS PARAMETERS
.PP
The following parameters are read from the parameter file:
.I " synt_interp,"
.I " synt_pwr,"
.I " synt_rc,"
.I " synt_order,"
.I " v_excit_method,"
.I " uv_excit_method,"
.I " synt_method."
.I " post_filt_num"
.I " post_filt_den"
.TP
.I synt_interp
.IP
If 
.I synt_interp
is "none", then no interpolation is performed on the reflection 
coefficients from the ANA file. 
.IP
.I synt_interp
is "pulse", then the the reflection coefficients from the ANA file are
interpolated at every pitch pulse.  
.IP
.I synt_interp
is "sample", then the the reflection coefficients from the ANA file are
interpolated at every sample.  
.TP
.I synt_pwr
.IP
If 
.I synt_pwr
is "rawpulse", then the power per pulse in the synthesized speech 
is made to match that of the analyzed speech.  
.IP
If
.I synt_pwr
is "lpcpulse", then the power per pulse in the synthesized speech 
is obtained from the LPC residual signal power.  
.TP
.I synt_rc
.IP
If
.I synt_rc
is "ana", then the reflection coefficients are taken as is from the 
input ANA file.  
.IP
If
.I synt_rc
is "sinx", then the reflection coefficients from the ANA file are
converted to autocorrelations and extrapolated, sin(x)/x windowing is
applied, and then the autocorrelations are converted back to reflection
coefficients for synthesis.  
.TP
.I synt_order
.IP
If 
.I synt_order 
= 0, then the order of the synthesis filter is determined by the 
ANA file header entries 
.I hd.ana->order_vcd
and
.I hd.ana->order_unvcd.
.IP
If 
.I synt_order
!= 0, then the order of the synthesis filter is
MIN(\fIsynt_order,hd.ana->order.vcd\fP) for voiced speech and 
MIN(\fIsynt_order,hd.ana->order.unvcd\fP) for unvoiced speech.  
.TP
.I v_excit_method
.IP
If
.I v_excit_method
is "impulse", a single impulse is used as the excitation function for 
each pitch pulse in the case of voiced speech.
.IP
If
.I v_excit_method
is "white", white noise is used as the excitation function for 
each input pitch pulse in the case of voiced speech.  The result is 
so-called "whispered speech".  
.TP
.I uv_excit_method
.IP
If
.I uv_excit_method
is "white", white noise is used as the excitation function for 
a single ANA frame in the case of unvoiced speech.  This is the only
excitation method for unvoiced speech supported at present.  
.TP
.I synt_method
.IP
IF 
.I synt_method 
is "psynch", pitch synchronous synthesis is performed.  This is 
the only supported value of 
.I synt_method
at present.  
.TP
.I post_filt_num - vector
.IP
A three element vector indicating the numerator of a pole-zero
post processing filter - for example {1.0, -1.0, 0.0}.
.TP
.I post_filt_den - vector
A three element vector indicating the denominator of a pole-zero
post processing filter - for example {1.0, -0.875, 0.0}.
.SH SPS HEADERS
.PP
The following parameters are read from the header of the input 
ANA file:  
.I hd.ana\->start, hd.ana\->nan, hd.ana\->frmlen, hd.ana\->order_vcd, 
.I hd.ana\->order_unvcd, hd.ana\->src_sf, hd.ana\->psynch,
.I hd.ana\->maxpulses, hd.ana\->maxraw, hd.ana\->maxlpc, hd.ana\->sincflag, 
.I hd.ana\->pre_emp.
.PP
.I
Synt
writes the usual values into the common part of the output file 
header.  
.PP
.I
Synt
writes the following header items in the output SD file:
.TP 
.I hd.sd\->equip
.IP
is set to NONE.
.TP
.I hd.sd\->max_value
.IP
is set to 0.
.TP
.I hd.sd\->scale
.IP
is set to 0.
.TP
.I hd.sd\->sf
.IP
is copied from the input header
.I hd.ana\->src_sf.
.TP
.I hd.sd\->src_sf
.IP
is copied from the input header
.I hd.ana\->src_sf.
.TP
.I hd.common\->refer
.IP
is copied from the 
.I hd.common\->refer 
of the input ANA file.  
.TP
.I hd.sd\->nchan
.IP
is set to 1.
.TP
.I hd.sd\->dcrem
.IP
is set to 0.
.TP
.I hd.sd\->synt_interp
.IP
is determined by the value in the parameter file.
.TP
.I hd.sd\->synt_pwr
.IP
is determined by the value in the parameter file.
.TP
.I hd.sd\->synt_rc
.IP
is determined by the value in the parameter file.
.TP
.I hd.sd\->synt_order
.IP
is determined by the value in the parameter file.
.TP
.I hd.sd\->v_excit_method
.IP
is determined by the value in the parameter file.
.TP
.I hd.sd\->uv_excit_method
.IP
is determined by the value in the parameter file.
.TP
.I hd.sd\->synt_method
is determined by the value in the parameter file.
.TP
.I hd.sd\->prefilter
is determined by the values in the parameter file (
.I post_filt_num
and
.I post_filter_den
).
.SH OPTIONS
The following options are supported:
.TP
.B \-d
This option should be specified if the synthesis very close to that of the 
dsp synthesizer needs to be performed. Otherwise, some pitch smoothing is
performed by the synthesizer.
.TP
.BI \-h " his_file"
If 
.I his_file
is specified,
.I synt
lists 

(i) the location at which the amplitude of the synthesized speech is likely
to be clipped,

(ii) pulse locations at which pitch doubling is noticed,

(iii)
pulse locations at which
some discontinuity in pulse duration estimates are noticed, and 

(iv) when 
.I synt_pwr
= RAWPULSES, pulse locations at which power matching does not succeed.
.TP
.BI \-x " debug_level"
If 
.I debug_level
is positive,
.I synt 
prints various information or debugging messages, on the standard error.
The messages are more verbose when
.I debug_level
is larger.  If
.I debug_level
is 0, no message is printed.
The default is 0.
.TP
.BI \-P " param_file"
uses the parameter file 
.I param_file
rather than the default which is
.I params.
.TP
.BI \-p " range"
Selects a subrange of points to be synthesized, where the start and end
points are defined with respect to the original SD file that is the
source of the input ANA file.  The range is specified using the format 
.I start\-end
or
.IR start:end .
Either 
.I start
or 
.I end
may be omitted, in which case the omitted parameter defaults respectively
to the start or end of the input ANA file.
.TP
.BI \-s
Write the ouput file as SHORT instead of FLOAT.
.SH FUTURE CHANGES
.PP
Modification will occur when we include silence detection in the overall
algorithm.
.SH SUBROUTINES CALLED:
.PP
extendrc,  fowrdfilt,  gauss,  getatal,  latticefilt,  preemp,  rand,  rctoc
smartsynt.
.SH SEE ALSO
ana (1-SPS), ptrack (1-SPS), SPS(5-SPS), ANA(5-SPS), SD(5-SPS).
.SH "BUGS"
.PP
None known.
.SH "AUTHOR"
.PP
S. Shankar Narayan; modified for SPS by Rodney Johnson and John Shore.