Position Paper for W3C Workshop on Internationalizing SSML The Usage of PartOfSpeech for Resolving M

1
Position Paper for W3C Workshop on
Internationalizing SSMLThe Usage of
Part-Of-Speech for Resolving Multiple
Pronunciations in SSML

• 2005. 11. 3.
• Myoung-Wan Koo and Du-Seong Chang
• KT/KAIT

2
Introduction

• Multiple pronunciation problem
• Same word but different pronunciations
• Newton /njut?n/ v.s. /nut?n/
• Same spelling but different pronunciations
  (homograph)
• refuse /r?'fjuz/ v.s. /'refjus/

lt?xml version"1.0" encoding"UTF-8"?gt ltlexicon
version"1.0" xmlns"http//www.w3.org/2005/01/pro
nunciation-lexicon" alphabet"ipa"
xmllang"en-GB"gt ltlexemegt
ltgraphemegtNewtonlt/graphemegt
ltphonemegtnjut?nlt/phonemegt ltphonemegtnut?nlt/ph
onemegt lt/lexemegt ltlexemegt
ltgraphemegtrefuselt/graphemegt ltphonemegt
r?'fjuz lt/phonemegt ltphonemegt'refjuslt/phoneme
gt lt/lexemegt lt/lexicongt
3
Multiple pronunciation in SSMLPLS

• SSML
• The Speech Synthesis Markup Language
  Specification Version 1.0
• Pronunciation information in SSML
• Phoneme element
• Lexicon element
• PLS
• Pronunciation Lexicon Specification Version 1.0
• Pronunciation information in PLS
• Phoneme element
• Prefer attribute
• They doesnt fully support the pronunciation
  lexicon for multiple pronunciations and
  agglutinative language.
• ? Part-Of-Speech information is needed

4
Pronunciation information in PLS (1/2)

• Pronunciation Lexicon Specification
• Version 1.0/Feb 2005/W3C Voice Browser Working
  Group
• It allow interoperable specification of
  pronunciation information for either ASR and TTS
  engines within voice browsing applications.
• It is expected to handle multiple pronunciation.
• Example of PLS

lt?xml version"1.0" encoding"UTF-8"?gt ltlexicon
version"1.0" xmlnshttp//www.w3.org/2005/01/pro
nunciation-lexicon alphabet"ipa"
xmllang"en-US"gt ltlexemegt
ltgraphemegttomatolt/graphemegt ltphonemegt
t?mei??oult/phonemegt lt/lexemegt lt/lexicongt
5
Pronunciation information in PLS (2/2)

• Prefer attribute of phoneme element
• Give one pronunciation high priority among
  pronunciation candidates.
• Effective in speech synthesis
• Only in multiple pronunciations for same
  orthography
• Not in homograph problem
• refuse verb/r?'fjuz/ v.s. noun/'refjus/
• No information for ASR systems.

lt?xml version"1.0" encoding"UTF-8"?gt ltlexicon
version"1.0" xmlns"http//www.w3.org/2005/01/pro
nunciation-lexicon" alphabet"ipa"
xmllang"en-GB"gt ltlexemegt
ltgraphemegtNewtonlt/graphemegt ltphoneme
prefer"true"gtnjut?nlt/phonemegt
ltphonemegtnut?nlt/phonemegt lt/lexemegt lt/lexicongt
6
Typical Korean TTS system structure
Structural Information
Morphemes, POS
Phonemes, POS
Phonemes, Prosody
Morphological Analyzer
Grapheme-to- Phoneme
Prosody Analysis
Waveform production
Text
Speech
7
POS for resolving multiple pronunciations

• POS information can reduce the overhead of
  resolving multiple pronunciations in ASR and TTS
  systems.
• The word refuse can have two different
  pronunciations depending on pos information.
• Proposal POS attribute

lt?xml version"1.0" encoding"UTF-8"?gt ltlexicon
version"1.0" xmlns"http//www.w3.org/2005/01/pro
nunciation-lexicon" alphabet"ipa"
xmllang"en-US"gt ltlexemegt
ltgraphemegtrefuselt/graphemegt ltphoneme
posverbgt r?'fjuz lt/phonemegt lt/lexemegt
ltlexemegt ltgraphemegtrefuselt/graphemegt
ltphoneme posnoungt'refjuslt/phonemegt
lt/lexemegt lt/lexicongt
8
POS information for LVCSR

• Large vocabulary continuous speech recognition of
  agglutinative language
• Basic unit is morpheme (pseudo-morpheme) for
  reducing the vocabulary size.
• Many homographs in the recognition dictionary.
• POS information help system to get a proper
  pronunciation in a dictionary as well as to
  resolve multiple pronunciations in some words.
• It reduce the search time since POS information
  could cut the wrong word connection in the first
  stage, not in the semantic interpretation stage.

9
Proposals

• Proposal 1 POS attribute of phoneme element
• Optional attribute
• Proposal 2 POS element
• Lexeme element contain optional POS elements.
• POS values language-specific
• Type allow vendor-specific POS type?
• Outstanding POS set Penn Treebank, Sejong
  project (Korean)

lt?xml version"1.0" encoding"UTF-8"?gt ltlexicon
version"1.0" xmlns"http//www.w3.org/2005/01/pro
nunciation-lexicon" alphabet"ipa"
xmllang"en-US"gt ltlexemegt
ltgraphemegtrefuselt/graphemegt ltphonemegt
r?'fjuz lt/phonemegt ltposgt verb lt/verbgt
lt/lexemegt lt/lexicongt
10
Conclusion

• No element or attribute for resolving multiple
  pronunciations
• In current SSML, PLS
• POS information
• can reduce the overhead of resolving multiple
  pronunciations in ASR and TTS systems.
• Can reduce the search time in a large vocabulary
  recognition system.
• Can be effective in agglutinative language.
• Proposals
• POS element
• POS attribute