u

script_nuance/tools/gather_transcription_from_wordsfile.pl

@@ -0,0 +1,35 @@
+print "Usage: perl $0 transcription_dir path_prefix output_file\n";
+print "Example: perl $0 /direct/datadigest/read_English_SG/gitm/transcription /direct/datadigest/read_English_SG/gitm/calls/ ~/gitm.list\n";
+
+use File::Find;
+use File::Copy;
+
+open CORPUS, ">$ARGV[2]" or die "Cannot open corpus file $ARGV[1] for write.\n";
+
+if ($ARGV[1] =~ m|/$|) # the parameter "path_prefix" is ended with /
+{
+    $prefix = $ARGV[1];
+}else
+{
+    $prefix = "$ARGV[1]/";
+}
+
+@dirs = ($ARGV[0]);
+
+find ( {wanted => \&wanted},
+       @dirs );
+
+sub wanted
+{
+    if (m|^([a-zA-Z0-9_]+)_(utt\d+)\.words$|)
+    {
+	$folder = $1;
+	$utt = $2;
+	$folder =~ m|^[A-Za-z]+(\d\d\d)|;
+	$group = $1; # usually it's 000, but not always. So $group need be extracted.
+	open WORDS, "$_" or die "Cannot open words file $_\n";
+	$words = <WORDS>;
+	chomp ($words);
+	print CORPUS "$prefix$group/$folder/${folder}_${utt}.ulaw\t$words\n";
+    }
+}