www.pudn.com > sphinx_recipe.zip > PruneToAligned.pl

#!/usr/bin/perl

# Given a transcript that has been force aligned, prune a list of file ids
# to cooresponding 1-1 with the transcripts.  This elimiantes any files
# that failed to get aligned.
#
# The utterances should have been listed in the same order in both files
# before the alignment and we assume this ordering in the pruning.
#
# Copyright 2006 by Keith Vertanen
#

use strict;

if ( @ARGV < 2 )
{
    print "$0  \n"; 
    exit(1);
}

my $transFile;
my $i;
my $pos;
my $idFile;
my $idLine;
my $id;
my $transLine;
my $transID;
my $idLineUpper;

($transFile, $idFile) = @ARGV;

open(IN_TRANS, $transFile);
open(IN_IDS, $idFile);

$transLine = uc();
$transLine =~ s/[\n\r]//g;

# There should be more lines in the id file than in the transcription file
while ($idLine = ) 
{
    $idLine =~ s/[\n\r]//g;
    $idLineUpper = uc($idLine);

    # Line from ID is like:
    # 	SI_TR_S/40O/40OC0209
    $id = substr($idLineUpper, rindex($idLineUpper, "/") + 1);
    
    # print "id = $id\n";
    
    # See if our current transcript line matches this ID
    if (index($transLine, "(" . $id . ")") != -1)
    {
	print $idLine . "\n";

	# Advance to the next transcript line
	$transLine = uc();
        $transLine =~ s/[\n\r]//g;
    }
}
close IN_TRANS;
close IN_IDS;