Early error detection on word level Gabriel Skantze and Jens Edlund {gabriel, edlund}@speech. kth. se Centre for Speech Technology Department of Speech, Music and Hearing KTH, Sweden TT Centre for Speech Technology
Overview • How do we handle errors in conversational humancomputer dialogue? • Which features are useful for error detection in ASR results? • Two studies on selected features: – Machine learning – Human subjects’ judgement
Error detection • Early error detection – Detect if a given recognition result contains errors – e. g. Litman, D. J. , Hirschberg, J. , & Swertz, M. (2000). • Late error detection – Feed back the interpretation of the utterance to the user (grounding) – Based on the user’s reaction to that feedback, detect errors in the original utterance – e. g. Krahmer, E. , Swerts, M. , Theune, T. & Weegels, M. E. (2001). • Error prediction – Detect that errors may occur later on in the dialogue – e. g. Walker, M. A. , Langkilde-Geary, I. , Wright Hastie, H. , Wright, J. , & Gorin, A. (2002).
Why early error detection? • ASR errors reflect errors in acoustic and language models. Why not fix them there? – Post-processing may consider systematic errors in the models, due to mismatched training and usage conditions. – Post-processing may help to pinpoint the actual problems in the models. – Post-processing can include factors not considered by the ASR, such as: • Prosody • Semantics • Dialogue history
Corpus collection Speaks ASR Listens Vocoder User I have the lawn on my right and a house with number two on my left i have the lawn on right is and a house with from two on left Reads Speaks Operator
Study I: Machine learning • • 4470 words 73. 2% correct (baseline) 4/5 training data, 1/5 test data Two ML algorithms tested – Transformation-based learning (µ-TBL) • Learn a cascade of rules that transforms the classification – Memory-based learning (Ti. MBL) • Simply store each training instance in memory • Compare the test instance to the stored instances and find the closest match
Features Group Feature Explanation Confidence Speech recognition word confidence score Lexical Word The word POS The part-of-speech for the word Length The number of syllables in the word Content Is it a content word? Prev. POS The part-of-speech for the previous word Next. POS The part-of-speech for the next word Prev. Word The previous word Prev. Dialogue. Act The dialogue act of the previous operator utterance Mentioned Is it a content word that has been mentioned previously by the operator in the discourse? Contextual Discourse
Results Feature set µ-TBL Ti. MBL Confidence 77. 3% 76. 0% Lexical 77. 5% 78. 0% Lexical + Contextual 81. 4% 82. 8% Lexical + Confidence 81. 3% 81. 0% Lexical + Confidence + Contextual 83. 9% 83. 2% Lexical + Confidence + Contextual + Discourse 85. 1% 84. 1% • Content-words: – Baseline: 69. 8%, µ-TBL: 87. 7%, Ti. MBL: 87. 0%
Rules learned by µ-TBL Transformation Rule TRUE > FALSE Confidence < 50 & Content = TRUE > FALSE Confidence < 60 & POS = Verb & Length = 2 TRUE > FALSE Confidence < 40 & POS = Adverb & Length = 1 TRUE > FALSE Confidence < 50 & POS = Adverb & Length = 2 TRUE > FALSE Confidence < 40 & POS = Verb & Length = 1 FALSE > TRUE Confidence > 40 & Mentioned = TRUE & POS = Noun & Length =2
Study II: Human error detection • • First 15 user utterances from 4 dialogues with high WER 50% of the words correct (baseline) 8 judges Features were varied for each utterance: – ASR information – Context information
Features No. Context No context. ASR output only. Previous. Context Previous utterance visible. Full. Context The dialogue history is given incrementally. Map. Context As Full. Context, with the addition of the map. No. Confidence Recognised string only. Confidence Recognised string, colour coded for word confidence. NBest. List As Confidence, but the 5 -best ASR result was given.
The judges’ interface Dialogue so far Correction field 5 -best list Grey scale reflect word confidence Utterance confidence
Results
Conclusions & Discussion • ML can be used for early error detection on word level, especially for content words. • Word confidence scores have some use. • Utterance context and lexical information improve the ML performance. • A rule-learning algorithm such as transformation-based learning can be used to pinpoint the specific problems. • N-best lists are useful for human subjects. How do we operationalise them for ML?
Conclusions & Discussion • The ML improved only slightly from the discourse context. – Further work in operationalising context for ML should focus on the previous utterance • The classifier should be tested together with a parser or keyword spotter to see if it can improve performance. • Other features should be investigated, such as prosody. These may improve performance further.
The End Thank you for your attention! Questions? TT Centre for Speech Technology
Скачать презентацию Early error detection on word level Gabriel Skantze
df858ccce454e55bee29d82b4d964d3d.ppt