1 #!/usr/bin/perl
2 use warnings; #sed replacement for -w perl parameter
3 #
4 # Copyright 2013 Daniel Povey
5 # Apache 2.0
7 if (@ARGV != 2) {
8 print STDERR "Usage: $0 <path-to-LDC2001S13> <path-to-output>\n";
9 print STDERR "e.g. $0 /export/corpora5/LDC/LDC2001S13 data/swbd_cellular1_train\n";
10 exit(1);
11 }
12 ($db_base, $out_dir) = @ARGV;
14 if (system("mkdir -p $out_dir")) {
15 die "Error making directory $out_dir";
16 }
18 open(CS, "<$db_base/doc/swb_callstats.tbl") || die "Could not open $db_base/doc/swb_callstats.tbl";
19 open(GNDR, ">$out_dir/spk2gender") || die "Could not open the output file $out_dir/spk2gender";
20 open(SPKR, ">$out_dir/utt2spk") || die "Could not open the output file $out_dir/utt2spk";
21 open(WAV, ">$out_dir/wav.scp") || die "Could not open the output file $out_dir/wav.scp";
23 @badAudio = ("40019", "45024", "40022");
25 while (<CS>) {
26 $line = $_ ;
27 @A = split(",", $line);
28 if (/$A[0]/i ~~ @badAudio) {
29 # do nothing
30 } else {
31 $wav = "sw_" . $A[0];
32 $spkr1= "sw_" . $A[1];
33 $spkr2= "sw_" . $A[2];
34 $gender1 = $A[3];
35 $gender2 = $A[4];
36 if ($A[3] eq "M") {
37 $gender1 = "m";
38 } elsif ($A[3] eq "F") {
39 $gender1 = "f";
40 } else {
41 die "Unknown Gender in $line";
42 }
43 if ($A[4] eq "M") {
44 $gender2 = "m";
45 } elsif ($A[4] eq "F") {
46 $gender2 = "f";
47 } else {
48 die "Unknown Gender in $line";
49 }
50 if (-e "$db_base/$wav.sph") {
51 $uttId = $spkr1 . "-swbdc_" . $wav ."_1";
52 if (!$spk2gender{$spkr1}) {
53 $spk2gender{$spkr1} = $gender1;
54 print GNDR "$spkr1"," $gender1\n";
55 }
56 print WAV "$uttId"," sph2pipe -f wav -p -c 1 $db_base/$wav.sph |\n";
57 print SPKR "$uttId"," $spkr1","\n";
59 $uttId = $spkr2 . "-swbdc_" . $wav ."_2";
60 if (!$spk2gender{$spkr2}) {
61 $spk2gender{$spkr2} = $gender2;
62 print GNDR "$spkr2"," $gender2\n";
63 }
64 print WAV "$uttId"," sph2pipe -f wav -p -c 2 $db_base/$wav.sph |\n";
65 print SPKR "$uttId"," $spkr2","\n";
66 } else {
67 print STDERR "Missing $db_base/$wav.sph\n";
68 }
69 }
70 }
72 close(WAV) || die;
73 close(SPKR) || die;
74 close(GNDR) || die;
75 if (system("utils/utt2spk_to_spk2utt.pl $out_dir/utt2spk >$out_dir/spk2utt") != 0) {
76 die "Error creating spk2utt file in directory $out_dir";
77 }
78 if (system("utils/fix_data_dir.sh $out_dir") != 0) {
79 die "Error fixing data dir $out_dir";
80 }
81 if (system("utils/validate_data_dir.sh --no-text --no-feats $out_dir") != 0) {
82 die "Error validating directory $out_dir";
83 }