Blame view
egs/fisher_callhome_spanish/s5/local/callhome_make_trans.pl
3.12 KB
8dcb6dfcb
first commit
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 |
#!/usr/bin/env perl
# Copyright 2014 Gaurav Kumar. Apache 2.0
use utf8;
use File::Basename;
($tmpdir)=@ARGV;
$trans="$tmpdir/callhome_train_transcripts.flist";
$reco="$tmpdir/callhome_reco2file_and_channel";
open(T, "<", "$trans") || die "Can't open transcripts file";
open(R, "|sort >$reco") || die "Can't open reco2file_and_channel file $!";
open(O, ">$tmpdir/callhome.text.1") || die "Can't open text file for writing";
open(G, ">$tmpdir/callhome_spk2gendertmp") || die "Can't open the speaker to gender map file";
binmode(O, ":utf8");
while (<T>) {
$file = $_;
m:([^/]+)\.txt: || die "Bad filename $_";
$call_id = $1;
print R "$call_id-A $call_id A
";
print R "$call_id-B $call_id B
";
open(I, "<$file") || die "Opening file $_";
binmode(I, ":iso88591");
#Now read each line and extract information
while (<I>) {
#136.37 138.10 B: Ah, bueno, mamita.
chomp;
my @stringComponents = split(":", $_, 2);
my @timeInfo = split(" ", $stringComponents[0]);
$stringComponents[1] =~ s/^\s+|\s+$//g ;
my $words = $stringComponents[1];
#Check number of components in this array
if ((scalar @stringComponents) >= 2) {
$start = sprintf("%06d", $timeInfo[0] * 100);
$end = sprintf("%06d", $timeInfo[1] * 100);
length($end) > 6 && die "Time too long $end in $file";
$side = "A";
if (index($timeInfo[2], "B") != -1) {
$side = "B";
}
$utt_id = "${call_id}-$side-$start-$end";
$speaker_id = "${call_id}-$side";
# All speakers are treated as male because speaker gender info
# is missing in this file
$gender = "m";
print G "$speaker_id $gender
" || die "Error writing to speaker2gender file";
$words =~ s|\[\[[^]]*\]\]||g; #removes comments
$words =~ s|\{laugh\}|\$laughter\$|g; # replaces laughter tmp
$words =~ s|\[laugh\]|\$laughter\$|g; # replaces laughter tmp
$words =~ s|\{[^}]*\}|\[noise\]|g; # replaces noise
$words =~ s|\[[^]]*\]|\[noise\]|g; # replaces noise
$words =~ s|\[/*([^]]*)\]|\[noise\]|g; # replaces end of noise
$words =~ s|\$laughter\$|\[laughter\]|g; # replaces laughter again
$words =~ s|\(\(([^)]*)\)\)|\1|g; # replaces unintelligible speech
$words =~ s|<\?([^>]*)>|\1|g; # for unrecognized language
$words =~ s|background speech|\[noise\]|g;
$words =~ s|background noise|\[noise\]|g;
$words =~ s/\[/larrow/g;
$words =~ s/\]/rarrow/g;
$words =~ s/[[:punct:]]//g;
$words =~ s/larrow/\[/g;
$words =~ s/rarrow/\]/g;
$words =~ s/[¿¡]//g;
$words =~ s/\h+/ /g; # horizontal whitespace characters
$words = lc($words);
print O "$utt_id $words
" || die "Error writing to text file";
}
}
close(I);
}
close(T);
close(R);
close(O);
close(G);
|