Hierarchical Neural Network Structures for Phoneme Recognition

  • Daniel Vasquez
  • Rainer Gruhn
  • Wolfgang Minker

Part of the Signals and Communication Technology book series (SCT)

Table of contents

  1. Front Matter
    Pages 1-15
  2. Daniel Vasquez, Rainer Gruhn, Wolfgang Minker
    Pages 1-6
  3. Daniel Vasquez, Rainer Gruhn, Wolfgang Minker
    Pages 7-30
  4. Daniel Vasquez, Rainer Gruhn, Wolfgang Minker
    Pages 31-48
  5. Daniel Vasquez, Rainer Gruhn, Wolfgang Minker
    Pages 49-59
  6. Daniel Vasquez, Rainer Gruhn, Wolfgang Minker
    Pages 61-101
  7. Daniel Vasquez, Rainer Gruhn, Wolfgang Minker
    Pages 103-117
  8. Daniel Vasquez, Rainer Gruhn, Wolfgang Minker
    Pages 119-122
  9. Back Matter
    Pages 0--1

About this book

Introduction

In this book, hierarchical structures based on neural networks are investigated for automatic speech recognition. These structures are evaluated on the phoneme recognition task where a  Hybrid Hidden Markov Model/Artificial Neural Network paradigm is used. The baseline hierarchical scheme consists of two levels each which is based on a Multilayered Perceptron. Additionally, the output of the first level serves as a second level input. The computational speed of the phoneme recognizer can be substantially increased by removing redundant information still contained at the first level output. Several techniques based on temporal and phonetic criteria have been investigated to remove this redundant information. The computational time could be reduced by 57% whilst keeping the system accuracy comparable to the baseline hierarchical approach.

Keywords

Artificial Neural Network HMM/ANN Hybrid Hidden Markov Model Multilayered Perceptron MLP TIMIT database articulatory attributes phoneme recognition phonetic decoder phonotactics spoken language dialogue systems

Authors and affiliations

  • Daniel Vasquez
    • 1
  • Rainer Gruhn
    • 2
  • Wolfgang Minker
    • 3
  1. 1., Institute of Information TechnologyUniversity of UlmUlmGermany
  2. 2.SVOX Deutschland GmbHUlmGermany
  3. 3., Institute of Information TechnologyUniversity of UlmUlmGermany

Bibliographic information

  • DOI https://doi.org/10.1007/978-3-642-34425-1
  • Copyright Information Springer Berlin Heidelberg 2013
  • Publisher Name Springer, Berlin, Heidelberg
  • eBook Packages Engineering
  • Print ISBN 978-3-642-34424-4
  • Online ISBN 978-3-642-34425-1
  • Series Print ISSN 1860-4862
  • About this book