From 18e1be0067bca38aacd31b1fb0e8addbde5bf629 Mon Sep 17 00:00:00 2001
From: Dan Povey <dpovey@gmail.com>
Date: Mon, 10 Mar 2014 01:53:55 +0000
Subject: [PATCH] sandbox/lid: data-prep script and run.sh fixes.

git-svn-id: https://svn.code.sf.net/p/kaldi/code/sandbox/language_id@3756 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8
---
 egs/lre/v1/local/make_lre03.pl |  4 ++--
 egs/lre/v1/local/make_lre05.pl | 10 ++++++++--
 egs/lre/v1/run.sh              |  2 +-
 3 files changed, 11 insertions(+), 5 deletions(-)

diff --git a/egs/lre/v1/local/make_lre03.pl b/egs/lre/v1/local/make_lre03.pl
index 50085d76b..b7b29f656 100755
--- a/egs/lre/v1/local/make_lre03.pl
+++ b/egs/lre/v1/local/make_lre03.pl
@@ -50,7 +50,7 @@ while($line = <DB>) {
   }
   $uttId = "lre03_${seg_id}";
   
-  print WAV "$uttId"," sph2pipe -f wav -p -c ${channel} $wav"." |\n";
+  print WAV "$uttId"," sph2pipe -f wav -p -c ${channel} $wav |\n";
   print UTT2SPK "$uttId $uttId\n";
   print UTT2LANG "$uttId $lang\n";
   print SPK2GEN "$uttId $gender\n";
@@ -90,7 +90,7 @@ for $set ("lid96d1", "lid96e1") {
 
       $wav = "$data_dir/$duration/$seg_id.sph";
       $uttId = "${set}_${seg_id}";
-      print WAV "$uttId"," sph2pipe -f wav -p -c 1 $wav"." |\n";
+      print WAV "$uttId"," sph2pipe -f wav -p -c 1 $wav |\n";
       print UTT2SPK "$uttId $uttId\n";
       print UTT2LANG "$uttId $lang\n";
       # Gender information is absent here, not outputting spk2gender file.
diff --git a/egs/lre/v1/local/make_lre05.pl b/egs/lre/v1/local/make_lre05.pl
index 650c63fbe..f77984c4d 100755
--- a/egs/lre/v1/local/make_lre05.pl
+++ b/egs/lre/v1/local/make_lre05.pl
@@ -33,7 +33,7 @@ while($line = <KEY>) {
   chomp($line);
   # If the line isn't a comment
   if (index($line, "#") == -1) {
-    ($fi, $lang, $conv_id, $chan, $test_cut) = split(" ", $line);
+    ($fi, $lang, $conv_id, $channel, $test_cut) = split(" ", $line);
     # Verify that we have only Indian English.
     if (not ($lang eq "IE")) {
       die "$db_ie contains non-Indian English utterances.";
@@ -43,7 +43,12 @@ while($line = <KEY>) {
     # This part of the corpus is only english.indian.
     $uttId = "lid05d1_$utt";
     $wav = $db_ie . $fi;
-    print WAV "$uttId"," sph2pipe -f wav -p -c $chan $wav |\n";
+    if (! -f $wav) {
+      print STDERR "No such file $wav (skipping)\n";
+      next;
+    }
+    $channel =~ tr/AB/12/;
+    print WAV "$uttId"," sph2pipe -f wav -p -c $channel $wav |\n";
     print UTT2SPK "$uttId $uttId\n";
     print UTT2LANG "$uttId english.indian\n";
   }
@@ -101,6 +106,7 @@ while($line = <KEY>) {
     }
 
     $uttId = "lid05e1_".$seg_id;
+    $channel =~ tr/AB/12/;
 
     print WAV "$uttId"," sph2pipe -f wav -p -c ${channel} $wav |\n";
     print UTT2SPK "$uttId $uttId\n";
diff --git a/egs/lre/v1/run.sh b/egs/lre/v1/run.sh
index 571242d54..df450f44b 100755
--- a/egs/lre/v1/run.sh
+++ b/egs/lre/v1/run.sh
@@ -51,7 +51,7 @@ rm foo
 ## HERE
 ##
 
-steps/make_mfcc.sh --mfcc-config conf/mfcc.conf --nj 40 --cmd "$train_cmd" \
+steps/make_mfcc.sh --mfcc-config conf/mfcc.conf --nj 100 --cmd "$train_cmd" \
   data/train exp/make_mfcc $mfccdir
 steps/make_mfcc.sh --mfcc-config conf/mfcc.conf --nj 40 --cmd "$train_cmd" \
   data/lre07 exp/make_mfcc $mfccdir