Unsupervised language model adaptation for broadcast news

Unsupervised language model adaptation for speech recognition is challenging, particularly for complicated tasks such the transcription of broadcast news (BN) data. This paper presents an unsupervised adaptation method for language modeling based on information retrieval techniques. The method is designed for the broadcast news transcription task where the topics of the audio data cannot be predicted in advance. Experiments are carried out using the LIMSI American English BN transcription system and the NIST 1999 BN evaluation sets. The unsupervised adaptation method reduces the perplexity by 7% relative to the baseline LM and yields a 2% relative improvement for a 10xRT system.

[1]  Marcello Federico,et al.  Bayesian estimation methods for n-gram language model adaptation , 1996, Proceeding of Fourth International Conference on Spoken Language Processing. ICSLP '96.

[2]  Renato De Mori,et al.  A Cache-Based Natural Language Model for Speech Recognition , 1990, IEEE Trans. Pattern Anal. Mach. Intell..

[3]  Jean-Luc Gauvain,et al.  Fast decoding for indexation of broadcast data , 2000, INTERSPEECH.

[4]  Jean-Luc Gauvain,et al.  Using information retrieval methods for language model adaptation , 2001, INTERSPEECH.

[5]  Dietrich Klakow,et al.  Language model adaptation using dynamic marginals , 1997, EUROSPEECH.

[6]  Jean-Luc Gauvain,et al.  The LIMSI Broadcast News transcription system , 2002, Speech Commun..

[7]  Marcello Federico,et al.  Efficient language model adaptation through MDI estimation , 1999, EUROSPEECH.