Indexing confusion networks for morph-based spoken document retrieval

In this paper, we investigate methods for improving the performance of morph-based spoken document retrieval in Finnish by extracting relevant index terms from confusion networks. Our approach uses morpheme-like subword units ("morphs") for recognition and indexing. This alleviates the problem of out-of-vocabulary words, especially with inflectional languages like Finnish. Confusion networks offer a convenient representation of alternative recognition candidates by aligning mutually exclusive terms and by giving the posterior probability of each term. The rank of the competing terms and their posterior probability is used to estimate term frequency for indexing. Comparing against 1-best recognizer transcripts, we show that retrieval effectiveness is significantly improved. Finally, the effect of pruning in recognition is analyzed, showing that when recognition speed is increased, the reduction in retrieval performance due to the increase in the 1-best error rate can be compensated by using confusion networks.

[1]  Andreas Stolcke,et al.  Finding consensus in speech recognition: word error minimization and other applications of confusion networks , 2000, Comput. Speech Lang..

[2]  Robert N. Goldberg,et al.  Standards in isothermal microcalorimetry (IUPAC Technical Report) , 2001 .

[3]  Philip T. Eubank,et al.  Consistency Tests for Binary VLE Data , 2000 .

[4]  Andreas Stolcke,et al.  SRILM - an extensible language modeling toolkit , 2002, INTERSPEECH.

[5]  W. Wagner,et al.  The IAPWS Formulation 1995 for the Thermodynamic Properties of Ordinary Water Substance for General and Scientific Use , 2002 .

[6]  Beth Logan,et al.  Word and sub-word indexing approaches for reducing the effects of OOV queries on spoken audio , 2002 .

[7]  Richard M. Stern,et al.  Integration of continuous speech recognition and information retrieval for mutually optimal performance , 1999 .

[8]  Mikko Kurimo,et al.  Unlimited vocabulary speech recognition with morph language models applied to Finnish , 2006, Comput. Speech Lang..

[9]  Juhani Aittamaa,et al.  Vapor-liquid equilibrium for binary system of thiophene + 2,2,4-trimethylpentane at 343.15 and 353.15 K and thiophene + 2-ethoxy-2-methylpropane at 333.15 and 343.15 K. , 2007 .

[10]  Richard Sproat,et al.  Lattice-Based Search for Spoken Utterance Retrieval , 2004, NAACL.

[11]  Milos Polednicek.  Developpement d'appareillages pour l'acquisition des donnees thermodynamiques dans les systemes d'interet environnemental et energetique , 2000 .

[12]  Kazuo Kojima,et al.  Evaluation of thermodynamic consistency of isobaric and isothermal binary vapor–liquid equilibrium data using the PAI test , 2004 .

[13]  G. M. Wilson,et al.  Vapor-Liquid Equilibrium. XI. A New Expression for the Excess Free Energy of Mixing , 1964 .

[14]  Mikko Kurimo,et al.  Using latent semantic indexing for morph-based spoken document retrieval , 2006, INTERSPEECH.

[15]  Mikko Kurimo,et al.  An evaluation of a spoken document retrieval baseline system in finish , 2004, INTERSPEECH.

[16]  Ellen M. Voorhees,et al.  The TREC Spoken Document Retrieval Track: A Success Story , 2000, TREC.

[17]  Kazuo Kojima,et al.  Thermodynamic consistency test of vapor-liquid equilibrium data: - Methanol ∼ water, benzene ∼ cyclohexane and ethyl methyl ketone ∼ water - , 1990 .

[18]  John P. O'Connell,et al.  A Generalized Method for Predicting Second Virial Coefficients , 1975 .

[19]  S. G. Penoncello,et al.  Thermodynamic Properties of Air and Mixtures of Nitrogen, Argon, and Oxygen From 60 to 2000 K at Pressures to 2000 MPa , 2000 .

[20]  David A. Hull.  Using statistical testing in the evaluation of retrieval experiments , 1993, SIGIR.

[21]  Karen Spärck Jones,et al.  The Cambridge University spoken document retrieval system , 1999, 1999 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings. ICASSP99 (Cat. No.99CH36258).

[22]  Janne Pylkkönen.  New pruning criteria for efficient decoding , 2005, INTERSPEECH.

[23]  Stephen Yerazunis,et al.  Vapor-liquid equilibrium determination by a new apparatus , 1964 .

[24]  Gerard Salton,et al.  A vector space model for automatic indexing , 1975, CACM.

[25]  Mikko Kurimo,et al.  To recover from speech recognition errors in spoken document retrieval , 2005, INTERSPEECH.

[26]  David Carmel,et al.  Spoken document retrieval from call-center conversations , 2006, SIGIR.

[27]  Kenney Ng,et al.  Subword-based approaches for spoken document retrieval , 2000, Speech Commun..

[28]  Juan Ortega,et al.  Vapor−Liquid Equilibria and Densities for Ethyl Esters (Ethanoate to Butanoate) and Alkan-2-ol (C3−C4) at 101.32 kPa , 1997 .

[29]  O. Redlich,et al.  Algebraic Representation of Thermodynamic Properties and the Classification of Solutions , 1948 .

[30]  Timothy J. Hazen,et al.  A comparison and combination of methods for OOV word detection and word confidence scoring , 2001, 2001 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.01CH37221).

[31]  Karen Spärck Jones,et al.  Effects of out of vocabulary words in spoken document retrieval (poster session) , 2000, SIGIR '00.

[32]  Mathias Creutz,et al.  Induction of the morphology of natural language : unsupervised morpheme segmentation with application to automatic speech recognition , 2006 .

[33]  Kazuo Kojima,et al.  Evaluation of thermodynamic consistency of isobaric and isothermal binary vapor–liquid equilibrium data using the PAI test II, alcohol + n-alkane, +aromatic, +cycloalkane systems , 2007 .