Large Vocabulary Continuous Speech Recognition - Berlin Chen

More documents

Recommendations

Info

WC: Language Model Look-ahead (1/2)• Language model probabilities incorporated as early inthe search as possible• Language model probability incorporated for computing of– Unigram Look-ahead– Bigram Look-aheadπv( a )= maxw ∈W( a )P( w v )• Anticipate the language modelprobabilities with the state hypothesis~QThrπn−1v1AC( a )( t,s;arc) = π ( a ) Q ( t,s;arc)() t= max⎡s, arc n−1v1= ⎢−n −1( v s arc )v11 , ,⎣w ∈W( a )P( w )⎤( as , arc) Q n ( t,s;arc ) × fACmax π1⎥⎦π( a)~Q−1v1n= maxw∈Wsarc( a )aP( w)Qv n1 −1( t,s;arc)( t,s;arc) < Thr ( t) ⇒ pruned!ACW( ) aImplementation of bigram look-ahead would be much difficult for time-conditionedthan word-conditioned search (why ?)Speech – Berlin Chen 26
WC: Language Model Look-ahead (2/2)• A Simple recursive function for calculating unigram LMlook-aheadvoid SpeechClass::Calculate_Word_Tree_Unigram(){if(Root==(struct Tree *) NULL) return;Do_Calculate_Word_Tree_Unigram(Root);}void SpeechClass::Do_Calculate_Word_Tree_Unigram(struct Tree *ptrNow){if(ptrNow==(struct Tree *) NULL) return;Do_Calculate_Word_Tree_Unigram(ptrNow->Brother);Do_Calculate_Word_Tree_Unigram(ptrNow->Child);if(ptrNow->Father!=(struct Tree *) NULL)if(ptrNow->Unigram > ptrNow->Father->Unigram)ptrNow->Father->Unigram=ptrNow->Unigram;}Speech – Berlin Chen 27
Page 1 and 2: Large Vocabulary ContinuousSpeech R
Page 3 and 4: Why LVCSR Difficult ? (2/2)• Maxi
Page 5 and 6: Classification of Approaches to LVC
Page 7 and 8: Word-conditioned vs. Time-condition
Page 9 and 10: Word-Conditioned Tree-Copy Search (
Page 11 and 12: Lexical/Phonetic Tree (2/2)• Reas
Page 13 and 14: Context-Dependent Phonetic Constrai
Page 15 and 16: Word-Conditioned Search (1/3)• Wo
Page 17 and 18: Word-Conditioned Search (3/3)Backtr
Page 19 and 20: WC: Implementation Issues (2/3)•
Page 21 and 22: WC: Pruning of Path Hypotheses (1/5
Page 23 and 24: WC: Pruning of Path Hypotheses (3/5
Page 25: WC: Pruning of Path Hypotheses (5/5
Page 29 and 30: WC: Acoustic Look-ahead (2/3)• Po
Page 31 and 32: More Details on Time-Conditioned Se
Page 33 and 34: TC: Decomposition of Search History
Page 35 and 36: TC vs. WC: Conflicting Expectations
Page 37 and 38: TC vs. A* search (2/2)- Operations
Page 39 and 40: Word Graph Rescoring (1/6)• Tree-
Page 41 and 42: Word Graph Rescoring (3/6)• If bi
Page 43 and 44: Word Graph Rescoring (5/6)• “Wo
Page 45: • Feature ExtractionConfiguration

Large Vocabulary Continuous Speech Recognition - Berlin Chen

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?