{ "paper_id": "O12-1023", "header": { "generated_with": "S2ORC 1.0.0", "date_generated": "2023-01-19T08:03:17.588391Z" }, "title": "Study on Keyword Spotting using Prosodic Attribute Detection for Conversational Speech", "authors": [ { "first": "Yu-Jui", "middle": [], "last": "Huang", "suffix": "", "affiliation": { "laboratory": "", "institution": "Information Engineering National Chia-Yi University", "location": {} }, "email": "" }, { "first": "Yin-Wei", "middle": [], "last": "Chung", "suffix": "", "affiliation": {}, "email": "" }, { "first": "Jui-Feng", "middle": [], "last": "Yeh", "suffix": "", "affiliation": {}, "email": "" } ], "year": "", "venue": null, "identifiers": {}, "abstract": "It is one of most essential issues to extract the keywords from conversational speech for understanding the utterances from speakers. This thesis aims at keyword spotting from spontaneous speech for keyword detecting. We proposed prosodic features that are used for keyword detection. The prosody words are segmented from speaker's utterance according to the pre-training decision tree. The supported vector machine is further used as the classifier to judge the prosody word is keyword or not. The prosody word boundary segmentation algorithm based on decision tree is illustrated. Besides the data driven feature, the knowledge obtained from the corpus observation is integrated in the decision tree. Finally, the keyword", "pdf_parse": { "paper_id": "O12-1023", "_pdf_hash": "", "abstract": [ { "text": "It is one of most essential issues to extract the keywords from conversational speech for understanding the utterances from speakers. This thesis aims at keyword spotting from spontaneous speech for keyword detecting. We proposed prosodic features that are used for keyword detection. The prosody words are segmented from speaker's utterance according to the pre-training decision tree. The supported vector machine is further used as the classifier to judge the prosody word is keyword or not. The prosody word boundary segmentation algorithm based on decision tree is illustrated. Besides the data driven feature, the knowledge obtained from the corpus observation is integrated in the decision tree. Finally, the keyword", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Abstract", "sec_num": null } ], "body_text": [ { "text": "in the focus part are extracted using prosody features by sported vector machine (SVM). According to the experimental results, we can find the proposed method outperform the phone verification approach especially in recall and accuracy. This shows the proposed approach is operative for keyword detecting.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "Keywords: Keyword spotting, prosodic feature, prosody word, spoken language. ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "( ) i i i P t t D E", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "( 1) ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "EQUATION", "cite_spans": [], "ref_spans": [], "eq_spans": [ { "start": 0, "end": 8, "text": "EQUATION", "ref_id": "EQREF", "raw_str": "Pi(t) i t i i bi ei i 2 2 ( )( ( ) ) , [ , ] ( ) i i i i e i i t b i i i e t b t t P t P t b e t t E \u00a6 \u00a6 ( 2) t 3 i P i 4 n 1 ( ) 2 i i t e b", "eq_num": "( 3)" } ], "section": "", "sec_num": null }, { "text": "EQUATION", "cite_spans": [], "ref_spans": [], "eq_spans": [ { "start": 0, "end": 8, "text": "EQUATION", "ref_id": "EQREF", "raw_str": "SVM (predict) +1 -1 5 1, 1, i i if T is semantic object T otherwise \u00ae ( 5) SVM (1) 01-10 i n i ij P i j 1 2 { , ,... } i i in i P P P PW Dur ij P i j Bi Ei _ i Syl N i _ ij Syl b i j _ ij Syl e i j", "eq_num": "(2)" } ], "section": "", "sec_num": null }, { "text": "Proceedings of the Twenty-Fourth Conference on Computational Linguistics and Speech Processing(ROCLING 2012)", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null } ], "back_matter": [ { "text": "[30] .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "annex", "sec_num": null }, { "text": "Proceedings of the Twenty-Fourth Conference on Computational Linguistics and Speech Processing (ROCLING 2012)", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "280-312. , , 2008", "sec_num": null } ], "bib_entries": { "BIBREF0": { "ref_id": "b0", "title": "A Study on Knowledge Source Integration for Candidate Rescoring in Automatic Speech Recognition", "authors": [ { "first": "J", "middle": [], "last": "Li", "suffix": "" }, { "first": "Y", "middle": [], "last": "Tsao", "suffix": "" }, { "first": "C", "middle": [ "H" ], "last": "Lee", "suffix": "" } ], "year": 2005, "venue": "ICASSP, IEEE International Conference", "volume": "1", "issue": "", "pages": "183--218", "other_ids": {}, "num": null, "urls": [], "raw_text": "J. Li, Y. Tsao and C.H. Lee, \"A Study on Knowledge Source Integration for Candidate Rescoring in Automatic Speech Recognition,\" ICASSP, IEEE International Conference, vol 1, pp837-840, 2005. [4] , , 11(2):183-218, 2010.", "links": null }, "BIBREF1": { "ref_id": "b1", "title": "Combining stochastic and linguistic language models for recognition of spontaneous speech", "authors": [ { "first": "E", "middle": [], "last": "Wieland", "suffix": "" }, { "first": "F", "middle": [], "last": "Gallwitz", "suffix": "" }, { "first": "H", "middle": [], "last": "Niemann", "suffix": "" } ], "year": 1996, "venue": "Proceedings of the IEEE Conference on Acoustics, Speech, and Signal Processing", "volume": "1", "issue": "", "pages": "423--426", "other_ids": {}, "num": null, "urls": [], "raw_text": "E. Wieland, F. Gallwitz, and H. Niemann. \"Combining stochastic and linguistic language models for recognition of spontaneous speech.\" In Proceedings of the IEEE Conference on Acoustics, Speech, and Signal Processing, vol.1, Atlanta, May, pp 423-426, 1996.", "links": null }, "BIBREF2": { "ref_id": "b2", "title": "Knowledge-based Parameters for HMM Speech Recognition", "authors": [ { "first": "N", "middle": [ "N" ], "last": "Bitar", "suffix": "" }, { "first": "C", "middle": [ "Y" ], "last": "Espy-Wilson", "suffix": "" } ], "year": 1996, "venue": "ICASSP", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "N. N. Bitar and C. Y. Espy-Wilson , \"Knowledge-based Parameters for HMM Speech Recognition,\" ICASSP 1996.", "links": null }, "BIBREF3": { "ref_id": "b3", "title": "A tutorial on hidden markov models and selected application in speech recognition", "authors": [ { "first": "L", "middle": [ "R" ], "last": "Rabiner", "suffix": "" } ], "year": 1989, "venue": "Proceedings of the IEEE", "volume": "77", "issue": "2", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "L. R. Rabiner, \"A tutorial on hidden markov models and selected application in speech recognition,\" Proceedings of the IEEE, vol.77, no. 2, Feb. 1989.", "links": null }, "BIBREF4": { "ref_id": "b4", "title": "Flexible Speech Understanding Based on Combined Key-Phrase Detection and Verification", "authors": [ { "first": "T", "middle": [], "last": "Kawahara", "suffix": "" }, { "first": "C", "middle": [ "H" ], "last": "Lee", "suffix": "" }, { "first": "B", "middle": [ "H" ], "last": "Juang", "suffix": "" } ], "year": 1998, "venue": "IEEE TRANSACTIONS ON SPEECH AND AUDIO PROCESSING", "volume": "6", "issue": "6", "pages": "558--568", "other_ids": {}, "num": null, "urls": [], "raw_text": "T. Kawahara, C.H. Lee, and B.H. Juang, \"Flexible Speech Understanding Based on Combined Key-Phrase Detection and Verification\", IEEE TRANSACTIONS ON SPEECH AND AUDIO PROCESSING, vol.6, NO. 6, pp.558-568, 1998.", "links": null }, "BIBREF5": { "ref_id": "b5", "title": "A Hidden Markov Model Based Keyword Recognition System", "authors": [ { "first": "R", "middle": [ "C" ], "last": "Rose", "suffix": "" }, { "first": "D", "middle": [ "B" ], "last": "Paul", "suffix": "" } ], "year": 1990, "venue": "Acoustics, Speech, and Signal Processing", "volume": "1", "issue": "", "pages": "129--132", "other_ids": {}, "num": null, "urls": [], "raw_text": "R. C. Rose, D. B. Paul, \"A Hidden Markov Model Based Keyword Recognition System\" Acoustics, Speech, and Signal Processing, ICASSP, vol.1, Page(s): 129 -132, 1990.", "links": null }, "BIBREF6": { "ref_id": "b6", "title": "A New Keyword Spotting Approach for Spontaneous Mandarin Speech", "authors": [ { "first": "P", "middle": [], "last": "Zhang", "suffix": "" }, { "first": "J", "middle": [], "last": "Han", "suffix": "" }, { "first": "J", "middle": [], "last": "Shao", "suffix": "" }, { "first": "Y", "middle": [], "last": "Yan", "suffix": "" } ], "year": 2006, "venue": "8th International Conference on", "volume": "1", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "P. Zhang, J. Han, J. Shao, Y. Yan, \"A New Keyword Spotting Approach for Spontaneous Mandarin Speech\" Signal Processing, 8th International Conference on vol.1, 2006.", "links": null }, "BIBREF7": { "ref_id": "b7", "title": "A New Keyword Spotting Approach\" Multimedia Computing and Systems, ICMCS, International Conference", "authors": [ { "first": "H", "middle": [], "last": "Bahi", "suffix": "" }, { "first": "N", "middle": [], "last": "Benati", "suffix": "" } ], "year": 2009, "venue": "", "volume": "", "issue": "", "pages": "77--80", "other_ids": {}, "num": null, "urls": [], "raw_text": "H. Bahi, N. Benati, \"A New Keyword Spotting Approach\" Multimedia Computing and Systems, ICMCS, International Conference , pp.77-80, 2009.", "links": null }, "BIBREF8": { "ref_id": "b8", "title": "Modeling out-of-vocabulary words for robust speech recognition", "authors": [ { "first": "I", "middle": [], "last": "Bazzi", "suffix": "" }, { "first": "J", "middle": [], "last": "Glass", "suffix": "" } ], "year": 2000, "venue": "Proc. ICSLP, Beijing", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "I. Bazzi and J. Glass, \"Modeling out-of-vocabulary words for robust speech recognition,\" Proc. ICSLP, Beijing, 2000.", "links": null }, "BIBREF9": { "ref_id": "b9", "title": "A new approach to utterance verification based on neighborhood information in model space", "authors": [ { "first": "H", "middle": [], "last": "Jiang", "suffix": "" }, { "first": "C", "middle": [ "H" ], "last": "Lee", "suffix": "" } ], "year": 2003, "venue": "IEEE Trans. Speech Audio Process", "volume": "11", "issue": "5", "pages": "425--434", "other_ids": {}, "num": null, "urls": [], "raw_text": "H. Jiang, C.H. Lee, \"A new approach to utterance verification based on neighborhood information in model space\", IEEE Trans. Speech Audio Process. 11(5), pp. 425-434, 2003.", "links": null }, "BIBREF10": { "ref_id": "b10", "title": "Bayesian Fusion of Confidence Measures for Speech Recognition", "authors": [ { "first": "T.-Y", "middle": [], "last": "Kim", "suffix": "" }, { "first": "H", "middle": [], "last": "Ko", "suffix": "" } ], "year": 2005, "venue": "IEEE SIGNAL PROCESSING LETTERS", "volume": "12", "issue": "12", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "T.-Y. Kim and H. Ko, \"Bayesian Fusion of Confidence Measures for Speech Recognition\", IEEE SIGNAL PROCESSING LETTERS, vol.12, NO. 12, Dec 2005.", "links": null }, "BIBREF11": { "ref_id": "b11", "title": "Improving the Performance of a Keyword Spotting System by Using Support Vector Machines", "authors": [ { "first": "Y", "middle": [], "last": "Benayed", "suffix": "" }, { "first": "D", "middle": [], "last": "Fohr", "suffix": "" }, { "first": "J", "middle": [ "P" ], "last": "Haton", "suffix": "" }, { "first": "G", "middle": [], "last": "Chollet", "suffix": "" } ], "year": 2003, "venue": "IEEE Auto Speech Recogniton and Understanding Workshop ASRU", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Y. BenAyed, D. Fohr, J. P. Haton, G. Chollet, \"Improving the Performance of a Keyword Spotting System by Using Support Vector Machines\", in IEEE Auto Speech Recogniton and Understanding Workshop ASRU, St, Thomas, U.S. Virgin islands, Dec 2003.", "links": null }, "BIBREF12": { "ref_id": "b12", "title": "Confidence measures for the Switchboard database", "authors": [ { "first": "R", "middle": [], "last": "Rose", "suffix": "" } ], "year": 1996, "venue": "Proc. of International Conference on Acoustics, Speech and Signal Processing", "volume": "", "issue": "", "pages": "511--514", "other_ids": {}, "num": null, "urls": [], "raw_text": "R. Rose, \"Confidence measures for the Switchboard database\", Proc. of International Conference on Acoustics, Speech and Signal Processing, pp.511-514, 1996.", "links": null }, "BIBREF13": { "ref_id": "b13", "title": "A Vector Space Modeling Approach to Spoken Language Identification", "authors": [ { "first": "H", "middle": [], "last": "Li", "suffix": "" }, { "first": "B", "middle": [], "last": "Ma", "suffix": "" }, { "first": "C", "middle": [ "H" ], "last": "Lee", "suffix": "" } ], "year": null, "venue": "Audio, Speech, and Language Processing", "volume": "15", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "H. Li, B. Ma, and C.H. Lee. \"A Vector Space Modeling Approach to Spoken Language Identification\", Audio, Speech, and Language Processing, IEEE Transactions on vol. 15,", "links": null } }, "ref_entries": { "TABREF0": { "text": "Proceedings of the Twenty-Fourth Conference on Computational Linguistics and Speech Processing(ROCLING 2012)", "num": null, "html": null, "type_str": "table", "content": "
Proceedings of the Twenty-Fourth Conference on Computational Linguistics and Speech Processing (ROCLING 2012)
Lee Case 5Key-Phrase Detection [27] SVMVerification[8] Tatsuya KawaharaChin-Hui SVM [4][5]
Case 6[9] Fujisaki Model (Keyword Detector) 3HPG (Prosodic Word Detection)9
(Keyword Detection) Case 7
[4][5][28][29][30]HPG
Case 8> (Pitch reset)Rose[10]HMM
Case 9
(filler)(Keyword spotting)Zhang[11]
Case 1(syllable)(prosodic word)
( (Dialogue system) (intonation phrase) (Pause) (1) 1 =0.04 Grouping, HPG)[4][5] Case 20.03HMM 0.05Bahi[12] (Hierarchical Prosodic Phrase ) (Spontaneous speech) (Speaking style) Bazzi 0.04
(2)(Grammar) (Real time) HMM[13](syllable, Syl)
Kawahara (prosodic phrase, PPh) ( 1) Kim[15] (prosodic phrase group, PG) (Verification) (prosodic word, PW) Lee C.H.[14] Case 3 (slope) i(Keyword extraction) (breath-group)
(Key-phrase detection) B1 B2 B3 B4(Key-phrase verification) B5(Sentence parsing) [16][17]
(sentence verification) (Incremental understanding) Haizhou Li, Bin Ma, and Chin-Hui Lee Case 4 Case 5 (Pitch Reset)B5 B1 Case 6[1] Case 7Case 8 [18] (Pitch Reset)Case 9
Charpter3 HPG[2]
B2
(Spoken Language Understanding, SLU)
29B
AuToBiConkie (Prosodic attribute)[20](Knowledge based) [19] (Pitch reset) POS[3]1HMM
deltaHMM
[4][5](Hierarchical Prosodic Phrase Grouping, HPG) (Prosodic word) Sridhar[21] HMM HMM 1 HPG
Erteschik-shir[22]
Case 1Ali1:[1] Wieland[23]>
Case 2Bi-gramBeam-search Viterbi
[24][6] Bitar
Case 3[25]HMM (Prosodic Attributes Extraction)
(Pitch)(Intensity)[7] Rabiner (Duration)1989
(HPG) (Prosodic Word Boundary) 2 Case 4 (Pitch reset)(Boundary Decision Tree) (Prosody word)[26]MFCC
" }, "TABREF1": { "text": "Proceedings of the Twenty-Fourth Conference on Computational Linguistics and Speech Processing(ROCLING 2012)", "num": null, "html": null, "type_str": "table", "content": "
iupper boundlower bound
iupper boundilower bound
upper boundilower bound
case 4pitch resetcase 8 pitch reset
i P1 n \u00a6 e i i t b( ) P t i( 4)
" }, "TABREF2": { "text": "Proceedings of the Twenty-Fourth Conference on Computational Linguistics and Speech Processing(ROCLING 2012)", "num": null, "html": null, "type_str": "table", "content": "
(False Positive, FP) 4 5 12 13 (c=10 g=16)(Keyword spotting) (True Negative, TN) (Semantic slot) (Speech act) 6 74.10% 52.90% [14] HTK forced alignment 69.19% HMM
01 02 03 04(Pragmatics) 4 4 5 11 12 13 4 5 5 (filler) (c=1 g=8) 4 5 11 12 13 ( ) Num i P PW i 15% (c=10 g=16) 3 5 6-9 12 (c=1 g=8) ( ) Dur i P PW i 3 5 6-9 12 (c=10 g=16) 4 5 6-8 12 Reference _ ( ) Dur Max i P PW i (c=1 g=8) 4 5 6-8 12 (c=10 g=16) Label + SVM _ ( ) Dur Min i P PW i(Topic) 6 77.42% 74.10% 75.83% 4 73.04% 74.90% accuracy 68% 71.58% 77.42%DA pair [23] 69.19% i n 68.45% 80.0% 1 i n Dur ij P \u00a6 j 77.73% recall 2 , ,..., Dur 1 Dur i i Max P P Erteschik-shir (Focus) 58.17% 4 52.94% 54.69% 51.25% 54.01% precision 70.22% { 68.45% 70.14% 49.5% 70.62% 58.17% 80% 1 2 { , ,..., Dur Dur i i Min P P} } Dur Dur i n P i n P
05Decision Tree + SVM ( ) i Dur PWi83.51%70.95%85.15% ( B E Pause PW i i i)
(2)SVM
06( Syl PW i)iSVM Syl N _ i
3
07 08 09 10 114: DA pairs 1 ( i Dur Syl 6 7 8 ) TP 2 ( ) i Dur Syl accuracy 100% i i TP FP TN FN 5: DA pairs (accuracy) 1 SVM TP TN 2 ( 6) TP precision TP FP 3: 3 ( ) i Dur Syl i 3 SVM SVM ( 7) Syl e Syl b 1 1 _ _ i i Syl e Syl b (precision) SVM 2 2 _ _ i i 3 3 _ _ i Syl e Syl b i accuracy precision recall 51%~58% 68%~80% 51%~59% 4 ( ) Dur Syl i 4 4 4 _ _ i i Syl e Syl b i (True TP recall TP FN ( 8) 4 5 12 13 (c=1 g=8) 83.38% 70.95% 75.33% Positive, TP) (recall) HPG 76%~83% 58%~71% 75%~85% ( ) i Pause PW i pause pause e b
12 13bpause (False Negative, FN) 2498 (True Positive, TP) SVM accuracy 850 precision 77.16% 57.83% 76.65% 58.42% Edinburgh Working Papers in Cognitive Science, 11:1-22, 1995. epause 13 12-13 13 660 2: (c=1 g=8) 4 5 6-8 12 (c=10 g=16) [2] N. Chater, M. Pickering, and D. Milward. \"What is incremental interpretation? \" recall 68.25% 75.22% 4 5 12 13 (c=1 g=8) 80.47% 65.02% ISCAS 1998. 75.33% 4 5 6-8 12 Feature-Based System for Automatic Phoneme Recognition in Continuous Speech,\" (c=10 g=16) 80.61% 63.00% 84.00% [1] Ali, J. Van der Spiegel, P. Mueller, G. Haentjens ,and J. Berman, \"An Acoustic-Phonetic 3 5 6-9 12 211 (3) 173 1061 211 58% (c=1 g=8) 82.45% 66.33% 85.15% 3 5 6-9 12 850 11 52 247 568 73 (1) SVM SVM 2 3-5% 58% 4 5 12 13 (c=10 g=16) 81.40% 65.41% 78.03% 4 5 11 12 13 83.51% 70.83% ( ) i pos PW i B i E 75.56% (c=1 g=8) (NSC 80% 10 8 4 5 11 12 13 (c=10 g=16) 81.35% 64.91% 77.48% 99-2221-E-415-006-MY3) . ( ) N Speech N
" } } } }