aliphn_to_textgrid.pl

#!/usr/bin/perl
# Copyright 2016 LeSpeech (Author: Xingyu Na)

if ($#ARGV != 2) {
  die "Usage: aliphn_to_textgrid.pl ali.1.phn data/lang/phones.txt ali.1\n" .
      "       ali.1.phn : generated by ali-to-phn with --write-lengths=true\n" .
      "       phones.txt: phone list generated when preparing lang\n" .
      "       ali.1     : output dir for TextGrid files\n";
}

$ali_fn = $ARGV[0];
$phn_fn = $ARGV[1];
$out_dir = $ARGV[2];

open PHN, $phn_fn || die "Cannot open $phn_fn!\n";

my %phn = ();

foreach (<PHN>) {
  chomp;
  @items = split /\s+/, $_;
  $phn{$items[1]} = $items[0];
}

close PHN;

open ALI, $ali_fn || die "Cannot open $ali_fn!\n";

mkdir $out_dir;
foreach (<ALI>) {
  chomp;
  @items = split /;/, $_;
  @subitems = split /\s+/, $items[0];
  $utt = $subitems[0];

  my @labs = ();
  @label = [$phn{$subitems[1]}, $subitems[2] * 0.01];
  push @labs, @label;
  my $len = $subitems[2] * 0.01;
  for ($i = 1; $i <= $#items; $i++) {
    @subitems = split /\s+/, $items[$i];
    push @labs, [$phn{$subitems[1]}, $subitems[2] * 0.01];
    $len = $len + $subitems[2] * 0.01;
  }
  open TXT, ">$out_dir/$utt.TextGrid" || die "Cannot write to $out_dir/$utt.TextGrid!\n";

  print TXT "File type = \"ooTextFile\"\n";
  print TXT "Object class = \"TextGrid\"\n\n";
  print TXT "xmin = 0\n";
  print TXT "xmax = $len\n";
  print TXT "tiers? <exists>\n";
  print TXT "size = 1\n";
  print TXT "item []:\n";
  print TXT "    item [1]:\n";
  print TXT "        class = \"IntervalTier\"\n";
  print TXT "        name = \"phn\"\n";
  print TXT "        xmin = 0\n";
  print TXT "        xmax = $len\n";
  $n = @labs;
  print TXT "        intervals: size = $n\n";
  $xmin = 0;
  $xmax = 0;
  for ($i = 0; $i <= $#labs; $i++) {
    $id = $i + 1;
    $xmin = $xmax;
    $xmax = $xmax + $labs[$i][1];
    print TXT "        intervals[$id]:\n";
    print TXT "            xmin = $xmin\n";
    print TXT "            xmax = $xmax\n";
    print TXT "            text = \"$labs[$i][0]\"\n";
  }

  close TXT;
}

close ALI;