umu.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Structured Language Models for Automatic Speech Transcription
Umeå University, Faculty of Science and Technology, Department of Computing Science.
2015 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

Speech recognition typically involves three types of models; an acoustic model, a phonetic dictionary and a language model. The primary purpose of the language model is to decide if a sentence is part of the language, and optionally how likely it is.

N-gram is a common type of language model which predicts upcoming words based on a series of prior words. While ecient a problem with this type of model is that it is not ideal for supporting long-distance dependencies. Tree-based models addresses this issue since they allow hierarchal constructs. The probabilistic lexicalized tree insertion grammar (PLTIG) is particularly interesting since it also limits the computational complexity of parsing sentences compared to similar methods such as tree-adjoining grammars of context-free grammars.

This thesis takes a closer look at the N-gram model and compares it with PLTIGs. Although PLTIGs tend to achieve higher accuracy the computational complexity is still very large in comparison.

Place, publisher, year, edition, pages
2015. , 48 p.
Series
UMNAD, 1006
National Category
Engineering and Technology
Identifiers
URN: urn:nbn:se:umu:diva-99646OAI: oai:DiVA.org:umu-99646DiVA: diva2:787377
External cooperation
CodeMill AB
Educational program
Master's Programme in Computing Science
Supervisors
Examiners
Available from: 2015-02-10 Created: 2015-02-10 Last updated: 2015-02-10Bibliographically approved

Open Access in DiVA

No full text

By organisation
Department of Computing Science
Engineering and Technology

Search outside of DiVA

GoogleGoogle Scholar

urn-nbn

Altmetric score

urn-nbn
Total: 1335 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf