Abstract
A set of features for word-level confidence estimation is developed. The features should be easy to implement and should require no additional knowledge beyond the information which is available from the speech recognizer and the training data. We compare a number of features based on a common scoring method, the normalized cross entropy. We also study different ways to combine the features. An artificial neural network leads to the best performance, and a recognition rate of 76% is achieved. The approach is extended not only to detect recognition errors but also to distinguish between insertion and substitution errors.
This is a preview of subscription content, log in via an institution to check access.
Preview
Unable to display preview. Download preview PDF.
References
L. Chase: Error-Responsive Feedback Mechanisms for Speech Recognition. Ph.D. Thesis, Carnegie Mellon University (1997).
B. Mison and R. Gopinath: Robust Confidence Annotation and Rejection for Continuous Speech Recognition Proc. IEEE ICASSP (2001) Vol. 1.
E. Eide and H. Gish and P. Jeanrenaud and A. Mielke: Understanding and Improving Speech Recognition Performance through the Use of Diagnostic Tools. Proc. IEEE ICASSP (1995) Vol. 1, 221–224.
S. Cox and R. C. Rose: Confidence Measures for the SWITCHBOARD Database. Proc. IEEE ICASSP (1996) Vol. 1, 511–514.
F. Wessel and K. Macherey and R. Schlüter: UsingWord Probabilities as Confidence Measures. Proc. IEEE ICASSP (1998) Vol. 1, 225–228.
W. Wahlster: Verbmobil: Foundations of Speech-to-Speech Translation. Springer (2000).
S. Steidl: Konfidenzbewertung von Worthypothesen. Student Thesis (in German), Chair for Pattern Recognition, University of Erlangen-Nürnberg (2001).
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2002 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Stemmer, G., Steidl, S., Nöth, E., Niemann, H., Batliner, A. (2002). Comparison and Combination of Confidence Measures. In: Sojka, P., Kopeček, I., Pala, K. (eds) Text, Speech and Dialogue. TSD 2002. Lecture Notes in Computer Science(), vol 2448. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-46154-X_25
Download citation
DOI: https://doi.org/10.1007/3-540-46154-X_25
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-44129-8
Online ISBN: 978-3-540-46154-8
eBook Packages: Springer Book Archive