Slide 1OutlineDefinitionSlide 4Can you determineHow it worksGrammarsTypes of GrammarsTag syntaxTaggrxml FileHARK FileHARK GrammarsSlide 14Dictionary FileTranscriptsAnalyzerReportPurposeWho uses these?ReferencesQuestions1Spoken Dialog SystemsSpeech RecognitionMarch 06, 2008COT 4810Albert Park2OutlineDefinitionHow it worksThe processPurpose of the use3DefinitionA Spoken dialog system is a dialog system delivered through voice. It has two essential components that do not exist in a text dialog system which are a speech recognizer and text-to-speech module.4DefinitionSpeech Recognition: It converts spoken words to machine-readable input. Also, attempts to identify the person speaking, as opposed to what is being said. Text-to-speech: A system that converts normal language into speech.5Can you determineThereTheirThey’re6How it worksAudioGrammarRun RecognitionAudioTranscriptRecognitionOutputAnalyzeAccuracyReport7GrammarsUses finite-state grammarsInitial stateInputsTransitionFinal state8Types of GrammarsHARKgrxml9Tag syntaxTag Type HARK grxmlWord word/tag N/ADot word ./tag word or $nterminal1 ./tag $nterminal2word <tag> tag </tag> wordNon-terminal$rule: $nterminal/tag <ruleref=URI> <tag> tag </tag> </ruleref>10TagText stringAny lengthNo spacesCase sensitive11grxml File<?xml version='1.0' encoding='utf-8' ?><!-- This is a comment --><grammar version='1.0' xml:lang='en-US' root='start'><rule id='start' scope='public'><one-of><!-- The list of names --> <item> JOHNATHAN <tag>id=1</tag> </item> <item> JOHN <tag>id=1</tag> </item> <item> MIKE <tag>id=2</tag> </item> <item> PETER <tag>id=3</tag> </item></one-of></rule></grammar>12HARK File<DICT> "include.hd";# Start here<Start> $START;$START:$FRIST;$FRIST:[$PRE] MY NAME IS $NAME [$END];$PRE:HELLO |HI;#List of names$NAME:JOHNATHAN/smith |JOHN/smith |MIKE/lee |PETER/joe;$END:NICE TO MEET YOU;13HARK Grammars<START>This is the initial state where the grammar starts from.<DICT>This acts as an include. When the speech recognition cannot determine the word it checks if the words exist in the dictionary file.14HARK Grammars$ For non-terminals[ ] Optional| or* more than one; end of operation# comments15Dictionary File#dictionary file>YES Y-EH-S>NO N-OW>KIANI KI-AH-NEE16TranscriptsThe actual speechSpeech into textTo compare17AnalyzerConverts speech to text.Compares and matches the voice and text.Shows the accuracy18ReportShows the correctShow substitutionShows insertionShows deletionShows errorsPerformance19PurposeLowering operating costReducing staffCustomer satisfaction20Who uses these?Automated telephone serviceCustomer serviceMilitaryPhone company21ReferencesLawrence R Rabiner, Biing-Hwang Juang, Fundamentals of Speech Recognition, Pearson Education, 2003Deborah Anna Dahl, Practical Spoken Dialog Speeches. Springer, 2004Wikipedia, Spoken Dialog System, Retrieved March 04, 2008 http://en.wikipedia.org/wiki/Spoken_dialogue_system22QuestionsWhat are two grammars for speech recognition?What is the reason we use speech recognition
View Full Document