Combined low level and high level features for out-of-vocabulary word detection

This paper addresses the issue of Out-Of-Vocabulary (OOV) word detection in Large Vocabulary Continuous Speech Recognition (LVCSR) systems. We propose a method inspired by confidence measures, that consists in analyzing the recognition system outputs in order to automatically detect errors due to OOV words. This method combines various features based on acoustic, linguistic, decoding graph and semantics. We evaluate separately each feature and we estimate their complementarity. Experiments are conducted on a large French broadcast news corpus from the ESTER evaluation campaign. Results show good performance in real conditions: the method obtains an OOV word detection rate of 43%-90% with 2.5%-17.5% of false detection. Index Terms: OOV word detection, confidence measures, speech recognition

[1]  Geoffrey Zweig,et al.  Confidence estimation, OOV detection and language ID using phone-to-word transduction and phone-level alignments , 2008, 2008 IEEE International Conference on Acoustics, Speech and Signal Processing.

[2]  James Glass,et al.  Modelling out-of-vocabulary words for robust speech recognition , 2002 .

[3]  James R. Glass,et al.  Modeling out-of-vocabulary words for robust speech recognition , 2000, INTERSPEECH.

[4]  Jie Zhu,et al.  OOV rejection algorithm based on class-fusion support vector machine for speech recognition , 2004, Proceedings of 2004 International Conference on Machine Learning and Cybernetics (IEEE Cat. No.04EX826).

[5]  Elmar Nöth,et al.  Semantic processing of out-of-vocabulary words in a spoken dialogue system , 1997, EUROSPEECH.

[6]  Bhiksha Raj,et al.  A boosting approach for confidence scoring , 2001, INTERSPEECH.

[7]  Hui Lin,et al.  OOV detection by joint word/phone lattice alignment , 2007, 2007 IEEE Workshop on Automatic Speech Recognition & Understanding (ASRU).

[8]  Paul Deléglise,et al.  Automatic Detection of Well Recognized Words in Automatic Speech Transcriptions , 2006, LREC.

[9]  Stephen Cox,et al.  A semantically-based confidence measure for speech recognition , 2000, INTERSPEECH.

[10]  Hui Sun,et al.  Using word confidence measure for OOV words detection in a spontaneous spoken dialog system , 2003, INTERSPEECH.

[11]  Georges Linarès,et al.  Local Methods for On-Demand Out-of-Vocabulary Word Retrieval , 2008, LREC.

[12]  Richard M. Schwartz,et al.  Automatic Detection Of New Words In A Large Vocabulary Continuous Speech Recognition System , 1989, HLT.

[13]  Guillaume Gravier,et al.  The ESTER phase II evaluation campaign for the rich transcription of French broadcast news , 2005, INTERSPEECH.

[14]  Kazuyo Tanaka,et al.  Detection of unknown words in large vocabulary speech recognition , 1993, EUROSPEECH.

[15]  Hermann Ney,et al.  Open vocabulary speech recognition with flat hybrid models , 2005, INTERSPEECH.

[16]  G. Boulianne,et al.  Out-of-vocabulary word modeling using multiple lexical fillers , 2001, IEEE Workshop on Automatic Speech Recognition and Understanding, 2001. ASRU '01..

[17]  Sheryl R. Young,et al.  Recognition Confidence Measures: Detection of Misrecognitions and Out- Of-Vocabulary Words , 1994 .