Segment boundary detection via class entropy measurements in connectionist phoneme recognition

This article investigates the possibility to use the class entropy of the output of a connectionist phoneme recogniser to predict time boundaries between phonetic classes. The rationale is that the value of the entropy should increase in proximity of a transition between two segments that are well modelled (known) by the recognition network since it is a measure of uncertainty. The advantage of this measure is its simplicity as the posterior probabilities of each class are available in connectionist phoneme recognition. The entropy and a number of measures based on differentiation of the entropy are used in isolation and in combination. The decision methods for predicting the boundaries range from simple thresholds to neural network based procedure. The different methods are compared with respect to their precision, measured in terms of the ratio between the number C of predicted boundaries within 10 or 20ms of the reference and the total number of predicted boundaries, and recall, measured as the ratio between C and the total number of reference boundaries.

[1]  T. Rejtar,et al.  A universal denoising and peak picking algorithm for LC-MS based on matched filtration in the chromatographic time domain. , 2003, Analytical chemistry.

[2]  Narada D. Warakagoda,et al.  A Noise Robust Multilingual Reference Recogniser Based on Speechdat(II) , 2000, INTERSPEECH.

[3]  Giampiero Salvi,et al.  SYNFACE - a talking face telephone , 2003, INTERSPEECH.

[4]  Steve Renals,et al.  Confidence measures from local posterior probability estimates , 1999, Comput. Speech Lang..

[5]  Giampiero Salvi.  Segment Boundaries in Low Latency Phonetic Recognition , 2005, NOLISP.

[6]  Mari Ostendorf,et al.  A Hierarchical Stochastic Model for Automatic Prediction of Prosodic Boundary Location , 1994, CL.

[7]  John-Paul Hosom.  Automatic phoneme alignment based on acoustic-phonetic modeling , 2002, INTERSPEECH.

[8]  Sharlene A. Liu,et al.  Landmark detection for distinctive feature-based speech recognition , 1996 .

[9]  James R. Glass.  A probabilistic framework for segment-based speech recognition , 2003, Comput. Speech Lang..

[10]  Jonas Beskow.  Trainable Articulatory Control Models for Visual Speech Synthesis , 2004, Int. J. Speech Technol..

[11]  Giampiero Salvi.  Dynamic behaviour of connectionist speech recognition with strong latency constraints , 2006, Speech Commun..

[12]  Giampiero Salvi.  Truncation error and dynamics in very low latency phonetic recognition , 2003, NOLISP.

[13]  C. Li,et al.  Detection of ECG characteristic points using wavelet transforms. , 1995, IEEE transactions on bio-medical engineering.

[14]  Kjell Elenius.  Two Swedish telephone speech databases , 1999 .

[15]  A. B.,et al.  SPEECH COMMUNICATION , 2001 .