Multi-level adaptive networks in tandem and hybrid ASR systems

P. Bell, P. Swietojanski, S. Renals

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In this paper we investigate the use of Multi-level adaptive networks (MLAN) to incorporate out-of-domain data when training large vocabulary speech recognition systems. In a set of experiments on multi-genre broadcast data and on TED lecture recordings we present results using of out-of-domain features in a hybrid DNN system and explore tandem systems using a variety of input acoustic features. Our experiments indicate using the MLAN approach in both hybrid and tandem systems results in consistent reductions in word error rate of 5-10% relative.
Original languageEnglish
Title of host publicationAcoustics, Speech and Signal Processing (ICASSP), 2013 IEEE International Conference on
PublisherInstitute of Electrical and Electronics Engineers (IEEE)
Pages6975-6979
Number of pages5
ISBN (Print)978-1-4799-0356-6
DOIs
Publication statusPublished - 2013

Keywords

  • error analysis
  • speech recognition
  • vocabulary
  • TED lecture recordings
  • hybrid ASR systems
  • multigenre broadcast data
  • multilevel adaptive networks
  • out-of-domain data
  • vocabulary speech recognition systems
  • word error rate
  • Acoustics
  • Adaptation models
  • Hidden Markov models
  • Neural networks
  • Speech
  • Speech recognition
  • Training
  • BBC
  • MLAN
  • TED
  • deep neural networks
  • hybrid
  • tandem

Fingerprint Dive into the research topics of 'Multi-level adaptive networks in tandem and hybrid ASR systems'. Together they form a unique fingerprint.

Cite this