Acoustic model optimisation for a call routing system

The paper presents work aimed at optimising acoustic models for the AutoSecretary call routing system. To develop the optimised acoustic models: (1) an appropriate phone set was selected and used to create a pronunciation dictionary, (2) various cepstral normalization techniques were investigated, (3) three South African corpora and multiple training data combinations were used to train the acoustic models, and, (4) model-space transformations were applied. Using an independent testing corpus, which contained proper names and South African language names, a named-language recognition accuracy of 95.11 % and proper name recognition accuracy of 93.31% were obtained.

[1]  Etienne Barnard,et al.  A Southern African corpus for multilingual name pronunciation , 2011 .

[2]  Stephen Cox,et al.  Some statistical issues in the comparison of speech recognition algorithms , 1989, International Conference on Acoustics, Speech, and Signal Processing,.

[3]  Mark J. F. Gales,et al.  Semi-tied covariance matrices for hidden Markov models , 1999, IEEE Trans. Speech Audio Process..

[4]  Thomas Niesler,et al.  The African Speech Technology Project: An Assessment , 2004, LREC.

[5]  F. Yates.  Contingency Tables Involving Small Numbers and the χ2 Test , 1934 .

[6]  Jean-Pierre Martens,et al.  Improving Proper Name Recognition by Adding Automatically Learned Pronunciation Variants to the Lexicon , 2010, LREC.

[7]  Vassilios Digalakis,et al.  Speaker adaptation using constrained estimation of Gaussian mixtures , 1995, IEEE Trans. Speech Audio Process..

[8]  Chris Schmandt,et al.  Putting people first: specifying proper names in speech interfaces , 1994, UIST '94.

[9]  Marelie H. Davel,et al.  ASR performance analysis of an experimental call routing system , 2009 .

[10]  Etienne Barnard,et al.  ASR corpus design for resource-scarce languages , 2009, INTERSPEECH.

[11]  Alta de Waal,et al.  Woefzela - An Open-Source Platform for ASR Data Collection in the Developing World , 2011, INTERSPEECH.

[12]  F. Bechet,et al.  Very large vocabulary proper name recognition for directory assistance , 2001, IEEE Workshop on Automatic Speech Recognition and Understanding, 2001. ASRU '01..

[13]  Ariadna Font Llitjós,et al.  Knowledge of language origin improves pronunciation accuracy of proper names , 2001, INTERSPEECH.

[14]  Thomas Niesler,et al.  Multi-accent acoustic modelling of South African English , 2012, Speech Commun..