.\" DO NOT MODIFY THIS FILE!  It was generated by help2man 1.47.3.
.TH SEQUITUR-G2P "1" "May 2016" "sequitur-g2p 0+r1668" "User Commands"
.SH NAME
sequitur-g2p \- grapheme-to-phoneme conversion tool
.SH SYNOPSIS
.B sequitur-g2p
[\fI\,OPTION\/\fR]... \fI\,FILE\/\fR...
.SH DESCRIPTION
Grapheme\-to\-Phoneme Conversion
.PP
Samples can be either in plain format (one word per line followed by
phonetic transcription) or Bliss XML Lexicon format.
.SH OPTIONS
.TP
\fB\-\-version\fR
show program's version number and exit
.TP
\fB\-h\fR, \fB\-\-help\fR
show this help message and exit
.TP
\fB\-p\fR FILE, \fB\-\-profile\fR=\fI\,FILE\/\fR
Profile execution time and store result in FILE
.TP
\fB\-R\fR, \fB\-\-resource\-usage\fR
Report resource usage execution time
.TP
\fB\-Y\fR, \fB\-\-psyco\fR
Use Psyco to speed up execution
.TP
\fB\-\-tempdir\fR=\fI\,PATH\/\fR
store temporary files in PATH
.TP
\fB\-t\fR FILE, \fB\-\-train\fR=\fI\,FILE\/\fR
read training sample from FILE
.TP
\fB\-d\fR FILE / N%, \fB\-\-devel\fR=\fI\,FILE\/\fR / N%
read held\-out training sample from FILE or use N% of
the training data
.TP
\fB\-x\fR FILE, \fB\-\-test\fR=\fI\,FILE\/\fR
read test sample from FILE
.TP
\fB\-\-checkpoint\fR
save state of training in regular time intervals. The
name of the checkpoint file is derived from \fB\-\-writemodel\fR.
.TP
\fB\-\-resume\-from\-checkpoint\fR=\fI\,FILE\/\fR
load checkpoint FILE and continue training
.TP
\fB\-T\fR, \fB\-\-transpose\fR
Transpose model, i.e. do phoneme\-to\-grapheme
conversion
.TP
\fB\-m\fR FILE, \fB\-\-model\fR=\fI\,FILE\/\fR
read model from FILE
.TP
\fB\-n\fR FILE, \fB\-\-write\-model\fR=\fI\,FILE\/\fR
write model to FILE
.TP
\fB\-\-continuous\-test\fR
report error rates on development and test set in each
iteration
.TP
\fB\-S\fR, \fB\-\-self\-test\fR
apply model to development set and report error rates
.TP
\fB\-s\fR l1,l2,r1,r2, \fB\-\-size\-constraints\fR=\fI\,l1\/\fR,l2,r1,r2
multigrams must have l1 ... l2 left\-symbols and r1 ...
r2 right\-symbols
.TP
\fB\-E\fR, \fB\-\-no\-emergence\fR
do not allow new joint\-multigrams to be added to the
model
.TP
\fB\-\-viterbi\fR
estimate model using maximum approximation rather than
true EM
.TP
\fB\-r\fR, \fB\-\-ramp\-up\fR
ramp up the model
.TP
\fB\-W\fR, \fB\-\-wipe\-out\fR
wipe out probabilities, retain only model structure
.TP
\fB\-C\fR, \fB\-\-initialize\-with\-counts\fR
initialize probabilities estimation by counting how
many times every graphone occurs in the training set,
disregarding possible overlaps
.TP
\fB\-i\fR MINITERATIONS, \fB\-\-min\-iterations\fR=\fI\,MINITERATIONS\/\fR
minimum number of EM iterations during training
.TP
\fB\-I\fR MAXITERATIONS, \fB\-\-max\-iterations\fR=\fI\,MAXITERATIONS\/\fR
maximum number of EM iterations during training
.TP
\fB\-\-eager\-discount\-adjustment\fR
re\-adjust discounts in each iteration
.TP
\fB\-\-fixed\-discount\fR=\fI\,D\/\fR
set discount to D and keep it fixed
.TP
\fB\-e\fR ENC, \fB\-\-encoding\fR=\fI\,ENC\/\fR
use character set encoding ENC
.TP
\fB\-P\fR, \fB\-\-phoneme\-to\-phoneme\fR
train/apply a phoneme\-to\-phoneme converter
.TP
\fB\-\-test\-segmental\fR
evaluate only at segmental level, i.e. do not count
syllable boundaries and stress marks
.TP
\fB\-B\fR FILE, \fB\-\-result\fR=\fI\,FILE\/\fR
store test result in table FILE (for use with bootlog
or R)
.TP
\fB\-a\fR FILE, \fB\-\-apply\fR=\fI\,FILE\/\fR
apply grapheme\-to\-phoneme conversion to words read
from FILE
.TP
\fB\-V\fR Q, \fB\-\-variants\-mass\fR=\fI\,Q\/\fR
generate pronunciation variants until \esum_i p(var_i)
>= Q (only effective with \fB\-\-apply\fR)
.TP
\fB\-\-variants\-number\fR=\fI\,N\/\fR
generate up to N pronunciation variants (only
effective with \fB\-\-apply\fR)
.TP
\fB\-f\fR FILE, \fB\-\-fake\fR=\fI\,FILE\/\fR
use a translation memory (read from sample FILE)
instead of a genuine model (use in combination with \fB\-x\fR
to evaluate two files against each other)
.TP
\fB\-\-stack\-limit\fR=\fI\,N\/\fR
limit size of search stack to N elements