Information retrieval methods for automatic speech recognition
Authors:
- 1 January 2010
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 5550-5553
Abstract
In this paper, we use information retrieval (IR) techniques to improve a speech recognition (ASR) system. The potential benefits include improved speed, accuracy, and scalability. Where conventional HMM-based speech recognition systems decode words directly, our IR-based system first decodes subword units. These are then mapped to a target word by the IR system. In this decoupled system, the IR serves as a lightweight, data-driven pronunciation model. Our proposed method is evaluated in the Windows Live Search for Mobile (WLS4M) task, and our best system has 12% fewer errors than a comparable HMM classifier. We show that even using an inexpensive IR weighting scheme (TF-IDF) yields a 3% relative error rate reduction while maintaining all of the advantages of the IR approach.Keywords
This publication has 9 references indexed in Scilit:
- Maximum mutual information multi-phone units in direct modeling2009
- A flat direct model for speech recognition2009
- A scalable method for voice search to nationwide business listings2009
- Hidden conditional random fields for phone classification2005
- Fast vocabulary-independent audio search using path-based graph indexing2005
- A direct adaptive method for faster backpropagation learning: the RPROP algorithm2002
- Subword-based approaches for spoken document retrievalSpeech Communication, 2000
- Statistical language modeling using the CMU-cambridge toolkit1997
- The zero-frequency problem: estimating the probabilities of novel events in adaptive text compressionIEEE Transactions on Information Theory, 1991