Changeset 2609 for Trunk/Scripts/Audio_scripts/AudioSegmentation/AudioBook/MissingWords.pm

Timestamp:

06/10/08 13:28:34 (7 months ago)

Author:

kmaclean

Message:

AudioSegmentation scripts - snapshot

Files:

• Trunk/Scripts/Audio_scripts/AudioSegmentation/AudioBook/MissingWords.pm (modified) (11 diffs)

Trunk/Scripts/Audio_scripts/AudioSegmentation/AudioBook/MissingWords.pm

@@ -112 +112 @@
-}
-
+=head2 getAudio 
+
+get audio object.
+
+=cut
+
+sub getAudio {
+        my ($self)= @_; 
+        my $audioBook = $self->{'audioBookObject'};
+        my $debug = $audioBook->getDebug;
+
+        $self->verifyMissingWordPronunciations();
+
+}
-
-=head2 verifyMissingWordPronunciations 
@@ -133 +147 @@
-        $dictionary->getNBestPronunciations("AudioBook/interim_files/MissingWords_alt",15); # uses Sequitor g2p to get top N pronunication vairations
-        $dictionary->createAltDict($originalDict,$altDict);     # merge & sort missing_words_alt and originalDict into altDict
-        # !!!!!! 
-        #$dictionary->validateAlternatePronunciations($originalDict,$altDict,$prompts);
-        $self->validateAlternatePronunciations($originalDict,$altDict,$prompts);
-        # !!!!!!
-}
-
@@ -154 +165 @@
-        my $audioBook = $self->{'audioBookObject'};
-        my $debug = $audioBook->getDebug;
-
+
-        my $dictionary = $self->{'dictionaryObject'};   
-        my $missing_words = $dictionary->getMissing_words();    
-
-
+
-        my $missingWordList = $self->_getMissingWordList($missing_words);
-        my %prompts;
@@ -173 +183 @@
-                        if ($missingWordList->{$word}) { # there is a missing word in this prompt line
-                                my $wavfilename = $promptID . "\.wav";
-                            
+
-                                push (@missingWordsPrompts,"$word:$promptID @line\n");
-                                print "\.";
-
-
+
+
+
+
+
+
-                        }
-                }
@@ -190 +204 @@
-                my @temp = split (/:/,$line);
-                my ($word) = shift(@temp); 
-
+#
-                if (defined($missingWordsPrompts{$word})) {
-                        my $temp = "$missingWordsPrompts{$word}\n";
@@ -256 +270 @@
-}
-
+=head2 _processMissingWordAudio 
+
+copy 
+
+=cut
+
+sub _processMissingWordAudio { # private 
+        my ($self,$promptID,$word,$startTime,$stopTime)= @_;
+        my $audioBook = $self->{'audioBookObject'};
+        my $debug = $audioBook->getDebug;       
+        my $chapter = $self->{'chapterObject'};
+        my $audio = $chapter->getchapterAudioObject(); 
+                
+        #copy("AudioBook/interim_files/wav/$wavfilename","AudioBook/interim_files/missingWordsFolder/$wavfilename");
+        my $fromFileName = "AudioBook/interim_files/wav/$promptID\.wav";
+        my $toFileName = "AudioBook/interim_files/missingWordsFolder/$promptID" . "-mw\.wav";
+        my $buffer = 2500000;
+        if ($startTime - $buffer > 0) {
+                $startTime = $startTime - $buffer;
+        } else {
+                $startTime = 0;
+        }
+        $stopTime = $stopTime + $buffer;        # htk will just padd the end of the file with silence if longer than full duration of file
+                
+        $audio->processMissingWordAudio($fromFileName, $startTime,$stopTime,$toFileName);
+        
+        return 1;
+}
+
-=head2 _forceAlignPromptLine 
-
@@ -261 +304 @@
-Sequitor G2P), in doing so, it picks the most likely pronunciation, thereby validating a Sequitor G2P with real audio.
-
-
+ (picks up only the first one...)
-
-=cut
@@ -273 +316 @@
-        open(ALIGNED_OUT,"$aligned_out") or confess ("cannot open $aligned_out file");
-        my (@phoneList,$gatherPhones);
+        my ($word_startTime,$word_stopTime);
-        while (my $line = <ALIGNED_OUT>) {
-                my @line = split(/\s/, $line);
@@ -279 +323 @@
-                        if ($recword eq $word) {
-                                $gatherPhones=1;
+                                $word_startTime = $startTime;
-                        } elsif ($gatherPhones) {
+                                $word_stopTime = $startTime; # can't search on "sp" because some words have not ending "sp" phone
-                                last;
-                        }
@@ -286 +332 @@
-                        if ($phone ne "sp"){
-                                push (@phoneList,$phone);
-
+ 
-                }
-        } 
-@phoneList
+(\@phoneList,$word_startTime,$word_stopTime)
-}
-
@@ -313 +359 @@
-}
-
-=head2 getAudio 
-
-get audio object.
-
-=cut
-
-sub getAudio {
-        my ($self,$segments)= @_;       
-        my $audioBook = $self->{'audioBookObject'};
-        my $debug = $audioBook->getDebug;
-        # structure $fileID,"$current_word $current_startTime $current_endTime $pause"
-        my $promptsWithTimes = $segments->getPromptsWithTimes();
-        open(MISSINGWORDS, "<AudioBook/interim_files/MissingWords") or confess ("cannot open AudioBook/output_files/prompts for output");
-        my %missingwords;
-        foreach my $word (<MISSINGWORDS>) {
-                $missingwords{$word}=1;
-        }
-        foreach my $line (@$promptsWithTimes) {
-                my $fileID = $$line[0];
-                my $promptWordsWithTime = $$line[1];
-                foreach my $line (@$promptWordsWithTime) {
-                        
-                }
-        }
-        
-        $self->{'segmentsObject'} = $segments;          
-}
-
-=head1 Change Log