Changeset 2593 for Trunk/Scripts/Audio_scripts/AudioSegmentation/AudioBook/Audio.pm

Timestamp:

05/26/08 14:11:03 (6 months ago)

Author:

kmaclean

Message:

AudioSegmentation scripts -add POD docs

Files:

• Trunk/Scripts/Audio_scripts/AudioSegmentation/AudioBook/Audio.pm (modified) (16 diffs)

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

@@ -1 +1 @@
-#! /usr/bin/perl
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
-package AudioBook::Audio;
-use strict;
@@ -48 +29 @@
-my $down_increment = -1;
-my $command;
-
-
-
+
+
+
+
+
+
+
+
+
+
+
-sub new {
-        my ($class,$super) = @_; 
@@ -66 +55 @@
-        return \%self;
-}    
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-sub _getSampleRate { # private
-        my ($self) = @_;        
@@ -79 +174 @@
-        return ($$details{sample_rate}); #sampling rate
-}
+
+=head3 _forceAlign 
+
+Performs "Forced Alignement" of the user submitted speech audio file.  This is the process of taking the text transcription 
+of the speech audio file and figuring you the time stamps for each of the words.  This helps determine where the pauses are.
+
+The time stamps are put in the "AudioBooks/interim_files/aligned.out" file 
+
+=cut
-
-sub     _forceAlign { # private
@@ -130 +234 @@
-}
-
+=head3 _processHViteOutput 
+
+This method reads the HVite output in the "aligned.out" file (generated by the _forceAlign method) and parses it into a 
+more user friendly format, and puts into the "AudioBooks/interim_files/htksegment_log" file. 
+
+=cut
+
-sub _processHViteOutput { # private
-        my ($self) = @_;        
@@ -170 +281 @@
-}
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
-
-sub _sentence_test { # private
@@ -328 +353 @@
-}
-
+=head3 _calculateStats 
+
+Calculates prompt statistics for Audio Segmenting summary i nthe AudioBook_Log as follows:
+
+ Sentence Length: min:aud0002: 11
+                 max:aud0003: 16
+
+ Prompt lines with more than max_sentence_length of 20 words:
+        none
+ 
+=cut
+
-sub _calculateStats { # Calculate min and max sentence
-        my ($self,$sentence_start,$sentence_end,$increment,$filename_prefix,$padded_fileid ) = @_;
@@ -357 +394 @@
-}
-
+=head3 _last_sentence 
+
+Special processing was requird for last prompt to segment properly.
+ 
+=cut
+
-sub _last_sentence { # private
-        my ($self, $aligned_words, $fileid, $sentence_start, $sentence_end) = @_;
@@ -389 +432 @@
-        _processAudio ($self, $startTime, $endTime,$padded_fileid);             
-}
+
+=head3 _processAudio 
+
+Create a segmented audio file from the original audio file using startTime and endTime generated from segmentation routines.
+ 
+=cut
-
-sub _processAudio { # private
@@ -413 +462 @@
-        }
-}
+
+=head2 verifySegments 
+
+Create a segmented audio file from the original audio file using startTime and endTime generated from segmentation routines.
+
+Second pass Forced Alignment on the segmented audio files generated from the segment method.  It tries to perform forced aligment 
+on each individual segmented audio file using the corresponded prompt line, to confirm that they match.  If they do not, it flags 
+the prompt for further editing. 
+ 
+=cut
-
-sub verifySegments { #public
@@ -450 +509 @@
-}
-
+=head3 _createMLFFile 
+
+Internal creates an MLF (HTK required "Multi-Label File") file for an individual prompt file.
+ 
+=cut
+
-sub _createMLFFile { # private
-        my ($self,$promptList)= @_;
@@ -462 +527 @@
-}
-
+=head3 _forceAlignSegment 
+
+Force Aligns an individual prompt lien with its corresponding speech audio file.
+ 
+=cut
+
-sub _forceAlignSegment {  # private
-        my ($self,$wavfilename)= @_;
@@ -470 +541 @@
-        $command = ("HVite -A -D -T 1 -a -b SENT-END -m -C $htk_files/wav_config -H $htk_files/models/macros -H $htk_files/models/hmmdefs -m -t 250.0 150.0 1000.0 -I AudioBook/interim_files/segment.mlf  -i AudioBook/interim_files/aligned.out AudioBook/interim_files/dict $htk_files/models/tiedlist AudioBook/interim_files/downsampled.wav > AudioBook/interim_files/logs/Segment_$wavfilename.log"); system($command) == 0 or confess "error: $command failed: $?";
-}
+
+=head3 _processHviteLog 
+
+Checks to make sure that the forced alignment process worked properly, if not, flags as an error.
+
+Checks for "No tokens survived to final node of network at beam" warnings in the HVite log file.
+ 
+=cut
-
-sub _processHviteLog {  # private
@@ -504 +583 @@
-}
-
-
-
-
+
+
+
+
+
+
-sub getFilename {
-        my $self = shift;
@@ -512 +594 @@
-}
-
+=item * getFilename_nopath()
+
+=cut
+
-sub getFilename_nopath {
-        my $self = shift;
@@ -517 +603 @@
-}
-
+=item * filename_prefix()
+
+=cut
+
-sub filename_prefix {
-        my $self = shift;
-        return $self->{"filename_prefix"} ;
-}
+
+=head1 Change Log    
+
+        2008/05/02 - 0.2 - Convert to class; major refacture; renamed from htksegment to Audio.pm
+        2007/06/12 - 0.1.2 - modularize code
+        2007.03.22 - 0.1 - created
+        
+=head1 Current Unresolved Issues:
+
+Warning:  Use of uninitialized value in numeric gt (>) "if ($sentence_length>$max_sentence_length) {"
+
+=head1 AUTHOR
+    
+Ken MacLean
+contact@voxforge.org
+      
+=head1 COPYRIGHT AND LICENSE       
+      
+Copyright (C) 2007 Ken MacLean
+   
+This program is free software; you can redistribute it and/or
+modify it under the terms of the GNU General Public License
+as published by the Free Software Foundation; either version 2
+of the License, or (at your option) any later version.
+   
+This program is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU General Public License for more details.
+    
+=cut
+    
-1;