Abstract

Accurate detection of dialogue acts is essential for understanding human conversations and to recognize emotions. This requires 1) the segmentation of human-human dialogs into turns, 2) the intra-turn segmentation into DA boundaries and 3) the classification of each segment according to a DA tag. Most dialogue act classification models approaches the problem of identifying the different DA segments within an utterance in separate fashion: first, DA boundary segmentation within an utterance was addressed with generative or discriminative approaches then, DA labels were assigned to such boundaries based on multi-classification. This paper, presents an effective approach to improve the accuracy of dialogue act recognition from speech signal by combining acoustic and linguistic features. This paper adopts the use of a silence removal algorithm based on Mahalanobis Distance for the segmentation of human-human dialogs into turns and proposes the keyword spotting feature to reduce the ambiguity of opinion vs. non-opinion statements and agreements vs. acknowledgements, occurs while classifying the dialogue acts.

References

• Y. Liu, A. Stolcke, E. Shriberg, and M. Harper. 2004. Comparing and combining generative and posterior probability models: Some advances in sentence boundary detection in speech. In Proceedings of the Conference on Empirical Methods in Natural Language Processing
• S. Quarteroni and G. Riccardi, "Dialog Act Classification in Human Human and Human Machine Conversations," in Proc. INTERSPEECH, 2010.
• Silvia Quarteroni, Alexei V. Ivanov, Giuseppe Riccardi, "Simultaneous Dialog Act Segmentation And Classification from Human-Human Spoken Conversations," IEEE 2011.
• G. Saha, Sandipan Chakroborty, Suman Senapati, "A New Silence Removal and Endpoint Detection Algorithm for Speech and Speaker Recognition Applications", Proc. Eleventh Conference on Speech Processing, IIT Khragpur 2005.
• M. Dinarelli, S. Quarteroni, S. Tonelli, A. Moschitti, and G. Riccardi,"Annotating spoken dialogs: from speech segments to dialog acts and frame semantics," in Proc. SRSL, 2009.
• J. Lafferty, A. McCallum, and F. Pereira, "Conditional random fields: Probabilistic models for segmenting and labelling sequence data," in Proc. ICML, 2001.
• P. Boersma, "Praat, a system for doing phonetics by computer," Glot International, vol. 5, no. 9/10, pp. 341–345, 2001. [Online]. Available: http://www. praat. org
• Atal, B. ; Rabiner, L. , "A pattern recognition approach to voiced-unvoiced-silence classification with applications to speech recognition" Acoustics, Speech, and Signal Processing [see also IEEE Transactions on Signal Processing], IEEE Transactions on , Volume: 24 , Issue: 3 , Jun 1976, Pages: 201 - 212.
• D. G. Childers, M. Hand, J. M. Larar, " Silent and Voiced/Unvoied/Mixed excitation(FourWay),Classification of Speech", IEEE Transaction on ASSP, Vol-37, No-11, pp. 1771-74, Nov 1989.
• A. Stolcke, K Ries, N. Coccaro, E. Shriberg, R. Bates, D. Jurafsky, P. Taylor, R. Martin, C. Van Ess-dykema, andM. Meteer, "Dialogue Act modeling and automatic tagging And recognition of conversational speech", Computational Linguistics vol. 26, 2000.
• Richard. O. Duda, Peter E. Hart, David G. Strok, "Pattern Classification", A Wiley Inter science publication, John Wiley & Sons, Inc, Second Edition, 2001.
• Sarma, V. ; Venugopal, D. , "Studies on pattern recognition approach to voiced-unvoiced-silence classification", Acoustics, Speech, and Signal Processing, IEEE International conference on ICASSP '78. ,Volume 3:April 1978, pages 1-4.