Navigation

← Previous Change
Next Change →

Dictionary.pm

Timestamp:

06/09/08 17:06:50 (6 months ago)

Author:

kmaclean

Message:

refacture to create Chapter, Segments & MissingWords? classes - snapshop

Files:

Trunk/Scripts/Audio_scripts/AudioSegmentation/AudioBook/Dictionary.pm (modified) (15 diffs)

Legend:

: Unmodified
: Added
: Removed
: Modified
: Copied
: Moved

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

r2604	r2606
1	1	#! /usr/bin/perl
2		$VERSION = 0.1;
	2	$VERSION = 0.1.1;
3	3
4	4	=head1 NAME
5	5
6		~~AudioBook::~~Dictionary - Pronunciation Dictionary processing
	6	Dictionary - Pronunciation Dictionary processing
7	7
8	8	=cut
9	9
10		package AudioBook::Dictionary;
	10	package AudioBook::Dictionary;
11	11	use strict;
12	12	use diagnostics;
…	…
31	31	sub new {
32	32	my ($class,$super) = @_;
33		my ~~%self;~~
34		~~$self{'htk_files'} = $super->{'htk_files'}~~;
35		~~$self{'g2p_model'} = $super->{'g2p_model'};~~
36		$self~~{'log'} = $super->{'log'};~~
37		bless(\%self,$class);
38		return \%self;
	33	my $self ={};
	34	bless($self,$class);
	35
	36	$self->{'audiobookObject'} = $super->getAudioBookObject();
	37
	38	return $self;
39	39	}
40	40
…	…
48	48
49	49	sub findOutOfVocabularyWords { # public
50		my ($self,$pronDict, $missing_words)= @_;
51		my $debug = $self->{"debug"};
52		my $htk_files = $self->{'htk_files'};
53
54		$self->{"pronDict"} = $pronDict;
55		$self->{"missing_words"} = $missing_words;
56
57		$command = ("HDMan -A -D -T 1 -g $htk_files/global.ded -m -w AudioBook/interim_files/wlist -i -l AudioBook/interim_files/dlog AudioBook/interim_files/dict $pronDict"); system($command) == 0 or confess "fullrun $command failed: $?";
	50	my ($self,$pronDict,$wlist,$missing_words)= @_;
	51	my $audioBook = $self->{'audiobookObject'};
	52	my $debug = $audioBook->getDebug();
	53	my $htk_files = $audioBook->getHtk_files();
	54	if ($debug) {
	55	$command = ("HDMan -A -D -T 1 -g $htk_files/global.ded -m -w $wlist -i -l AudioBook/interim_files/dlog AudioBook/interim_files/dict $pronDict"); system($command) == 0 or confess "fullrun $command failed: $?";
	56	} else {
	57	$command = ("HDMan -T 1 -g $htk_files/global.ded -m -w $wlist -i -l AudioBook/interim_files/dlog AudioBook/interim_files/dict $pronDict"); system($command) == 0 or confess "fullrun $command failed: $?";
	58	}
58	59	my $dlog1 = "AudioBook/interim_files/logs/dlog1";
59	60	$command = ("mv AudioBook/interim_files/dlog $dlog1"); print "cmd:$command\n" if $debug; system($command);
…	…
82	83	close DLOG;
83	84	close MISSINGWORDS;
	85
	86	$self->{"pronDict"} = $pronDict;
	87	$self->{"missing_words"} = $missing_words;
84	88	return $missingwordsfound;
85	89	}
86	90
87		=head2 get~~Recommended~~Pronunciations
	91	=head2 getG2PPronunciations
88	92
89	93	Use the Sequitor g2p script (Python) to generate pronunications for out-of-vocabulary words
…	…
91	95	=cut
92	96
93		sub get~~Recommended~~Pronunciations { # public
	97	sub getG2PPronunciations { # public
94	98	my ($self,$missing_words_out)= @_;
95		my $debug = $self->{'debug'};
96		my $model = $self->{'g2p_model'};
	99	my $audioBook = $self->{'audiobookObject'};
	100	my $debug = $audioBook->getDebug;
	101	my $model = $audioBook->getG2p_model();
97	102	my $missing_words = $self->{'missing_words'};
98		my $log = $self->{'log'};
99
100		$self->{"missing_words_out"} = $missing_words_out;
101
	103	my $log = $audioBook->getLog();
	104
102	105	my @missingwords = `export PYTHONPATH=/usr/local/lib64/python2.4/site-packages &&
103	106	g2p.py --model $model --apply $missing_words`;
…	…
135	138	close MISSINGWORDSOUT;
136	139	close LOG;
	140
	141	$self->{"missing_words_out"} = $missing_words_out;
137	142	return 1;
138	143	}
…	…
147	152	=cut
148	153
149		sub getAlternatePronunciations { # public
150		my ($self,$missing_words_alt, $numberOfPronunciations)= @_;
151		my $debug = $self->{'debug'};
152		my $model = $self->{'g2p_model'};
	154	sub getNBestPronunciations { # public
	155	my ($self,$missing_words_alt,$numberOfPronunciations)= @_;
	156	my $audioBook = $self->{'audiobookObject'};
	157	my $debug = $audioBook->getDebug;
	158	my $model = $audioBook->getG2p_model();
153	159	my $missing_words = $self->{'missing_words'};
154		my $log = $self->~~{'log'}~~;
	160	my $log = $self->getLog();
155	161
156	162	$self->{"missing_words_alt"} = $missing_words_alt;
…	…
188	194	sub validateAlternatePronunciations { # public
189	195	my ($self,$originalDict,$altDict,$prompts)= @_;
190		my $debug = $self->{'debug'};
	196	my $audioBook = $self->{'audiobookObject'};
	197	my $debug = $audioBook->getDebug;
191	198	my $missing_words = $self->{'missing_words'};
192	199
…	…
299	306	sub _forceAlignPromptLine { # private
300	307	my ($self,$altDict,$word,$promptID,$promptLine)= @_;
301
	308	my $audioBook = $self->{'audiobookObject'};
	309	my $debug = $audioBook->getDebug;
	310
302	311	my ($aligned_out, $log) = AudioBook::Audio->forceAlign($self, $promptID, $promptLine, $altDict);
303	312	open(ALIGNED_OUT,"$aligned_out") or confess ("cannot open $aligned_out file");
304	313	my (@phoneList,$gatherPhones);
305		~~#print "_forceAlignPromptLine:word=$word\n";~~
306	314	while (my $line = <ALIGNED_OUT>) {
307	315	my @line = split(/\s/, $line);
…	…
331	339	sub _getMissingWordList { # private
332	340	my ($self,$missing_words)= @_;
	341	my $audioBook = $self->{'audiobookObject'};
	342	my $debug = $audioBook->getDebug;
	343
333	344	my %missingWordList;
334	345	open(MISSINGWORDS,"$missing_words") or confess ("cannot open $missing_words file");
…	…
349	360	=cut
350	361
351		sub createAltDict { #
	362	sub createAltDict {
352	363	my ($self,$originalDict,$altDict)= @_;
	364	my $audioBook = $self->{'audiobookObject'};
	365	my $debug = $audioBook->getDebug;
353	366	my $missing_words_alt = $self->{"missing_words_alt"};
354
	367
355	368	open(MISSINGWORDSALT,"$missing_words_alt") or confess ("cannot open $missing_words_alt file");
356	369	my @missingWordAlt;
…	…
382	395	=head3 Note
383	396
384		The pronunciations generated by the Sequitor G2P scripts ~~need to~~ be manually reviewed before any new pronunciations
	397	The pronunciations generated by the Sequitor G2P scripts should be manually reviewed before any new pronunciations
385	398	are added to the pronunciation dictionary.
386	399
…	…
389	402	sub updatePronDict { # public
390	403	my ($self, $pronDictionary)= @_;
391		my $debug = $self->{"debug"};
	404	my $audioBook = $self->{'audiobookObject'};
	405	my $debug = $audioBook->getDebug;
392	406	my $missing_words_out = $self->{"missing_words_out"};
393	407
…	…
396	410	open(DICTIONARY,">>$pronDictionary") or confess ("cannot open $pronDictionary file");
397	411	open(MISSINGWORDSOUT,"<$missing_words_out") or confess ("cannot open $missing_words_out file");
398		print "adding the following words to VoxForge Pronunciation Dictionary\n";
	412	#print "adding the following words to VoxForge Pronunciation Dictionary\n";
399	413	while (my $line = <MISSINGWORDSOUT>) {
400	414	next if ($line =~ /^(\s)*$/); # skip blank lines
401	415	if ($line =~ /\w/) {
402		print $line;
	416	#print $line;
403	417	print DICTIONARY $line;
404	418	}
…	…
424	438	=head1 Change Log
425	439
	440	2008/06/09 - 0.1.1 - refacture to create Chapter, Segments & MissingWords classes
426	441	2008.05.05 - 0.1 - created
427	442

Dev Site (Scripts)

Navigation

Changeset 2606 for Trunk/Scripts/Audio_scripts/AudioSegmentation/AudioBook/Dictionary.pm

Legend:

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

Download in other formats: