Matches in SemOpenAlex for { <https://semopenalex.org/work/W2146168991> ?p ?o ?g. }
- W2146168991 endingPage "151" @default.
- W2146168991 startingPage "142" @default.
- W2146168991 abstract "Under-resourced speech recognizers may benefit from data in languages other than the target language. In this paper, we report how to boost the performance of an Afrikaans automatic speech recognition system by using already available Dutch data. We successfully exploit available multilingual resources through (1) posterior features, estimated by multilayer perceptrons (MLP) and (2) subspace Gaussian mixture models (SGMMs). Both the MLPs and the SGMMs can be trained on out-of-language data. We use three different acoustic modeling techniques, namely Tandem, Kullback–Leibler divergence based HMMs (KL-HMM) as well as SGMMs and show that the proposed multilingual systems yield 12% relative improvement compared to a conventional monolingual HMM/GMM system only trained on Afrikaans. We also show that KL-HMMs are extremely powerful for under-resourced languages: using only six minutes of Afrikaans data (in combination with out-of-language data), KL-HMM yields about 30% relative improvement compared to conventional maximum likelihood linear regression and maximum a posteriori based acoustic model adaptation." @default.
- W2146168991 created "2016-06-24" @default.
- W2146168991 creator A5037750159 @default.
- W2146168991 creator A5047712456 @default.
- W2146168991 creator A5076409146 @default.
- W2146168991 creator A5081808619 @default.
- W2146168991 date "2014-01-01" @default.
- W2146168991 modified "2023-10-03" @default.
- W2146168991 title "Using out-of-language data to improve an under-resourced speech recognizer" @default.
- W2146168991 cites W102093577 @default.
- W2146168991 cites W1526516369 @default.
- W2146168991 cites W1534441687 @default.
- W2146168991 cites W1549576988 @default.
- W2146168991 cites W1574284142 @default.
- W2146168991 cites W1583811137 @default.
- W2146168991 cites W1965555277 @default.
- W2146168991 cites W1968147312 @default.
- W2146168991 cites W1994126409 @default.
- W2146168991 cites W2002342963 @default.
- W2146168991 cites W2033436836 @default.
- W2146168991 cites W2090764203 @default.
- W2146168991 cites W2108776455 @default.
- W2146168991 cites W2112032806 @default.
- W2146168991 cites W2123798005 @default.
- W2146168991 cites W2128695877 @default.
- W2146168991 cites W2147276082 @default.
- W2146168991 cites W2151402979 @default.
- W2146168991 doi "https://doi.org/10.1016/j.specom.2013.01.007" @default.
- W2146168991 hasPublicationYear "2014" @default.
- W2146168991 type Work @default.
- W2146168991 sameAs 2146168991 @default.
- W2146168991 citedByCount "28" @default.
- W2146168991 countsByYear W21461689912013 @default.
- W2146168991 countsByYear W21461689912014 @default.
- W2146168991 countsByYear W21461689912015 @default.
- W2146168991 countsByYear W21461689912016 @default.
- W2146168991 countsByYear W21461689912017 @default.
- W2146168991 countsByYear W21461689912018 @default.
- W2146168991 countsByYear W21461689912019 @default.
- W2146168991 countsByYear W21461689912020 @default.
- W2146168991 countsByYear W21461689912021 @default.
- W2146168991 crossrefType "journal-article" @default.
- W2146168991 hasAuthorship W2146168991A5037750159 @default.
- W2146168991 hasAuthorship W2146168991A5047712456 @default.
- W2146168991 hasAuthorship W2146168991A5076409146 @default.
- W2146168991 hasAuthorship W2146168991A5081808619 @default.
- W2146168991 hasBestOaLocation W21461689912 @default.
- W2146168991 hasConcept C105795698 @default.
- W2146168991 hasConcept C137293760 @default.
- W2146168991 hasConcept C138885662 @default.
- W2146168991 hasConcept C14999030 @default.
- W2146168991 hasConcept C154945302 @default.
- W2146168991 hasConcept C155635449 @default.
- W2146168991 hasConcept C204321447 @default.
- W2146168991 hasConcept C207390915 @default.
- W2146168991 hasConcept C23224414 @default.
- W2146168991 hasConcept C28490314 @default.
- W2146168991 hasConcept C32834561 @default.
- W2146168991 hasConcept C33923547 @default.
- W2146168991 hasConcept C41008148 @default.
- W2146168991 hasConcept C41895202 @default.
- W2146168991 hasConcept C49781872 @default.
- W2146168991 hasConcept C50644808 @default.
- W2146168991 hasConcept C60908668 @default.
- W2146168991 hasConcept C61224824 @default.
- W2146168991 hasConcept C61328038 @default.
- W2146168991 hasConcept C91863865 @default.
- W2146168991 hasConcept C9810830 @default.
- W2146168991 hasConceptScore W2146168991C105795698 @default.
- W2146168991 hasConceptScore W2146168991C137293760 @default.
- W2146168991 hasConceptScore W2146168991C138885662 @default.
- W2146168991 hasConceptScore W2146168991C14999030 @default.
- W2146168991 hasConceptScore W2146168991C154945302 @default.
- W2146168991 hasConceptScore W2146168991C155635449 @default.
- W2146168991 hasConceptScore W2146168991C204321447 @default.
- W2146168991 hasConceptScore W2146168991C207390915 @default.
- W2146168991 hasConceptScore W2146168991C23224414 @default.
- W2146168991 hasConceptScore W2146168991C28490314 @default.
- W2146168991 hasConceptScore W2146168991C32834561 @default.
- W2146168991 hasConceptScore W2146168991C33923547 @default.
- W2146168991 hasConceptScore W2146168991C41008148 @default.
- W2146168991 hasConceptScore W2146168991C41895202 @default.
- W2146168991 hasConceptScore W2146168991C49781872 @default.
- W2146168991 hasConceptScore W2146168991C50644808 @default.
- W2146168991 hasConceptScore W2146168991C60908668 @default.
- W2146168991 hasConceptScore W2146168991C61224824 @default.
- W2146168991 hasConceptScore W2146168991C61328038 @default.
- W2146168991 hasConceptScore W2146168991C91863865 @default.
- W2146168991 hasConceptScore W2146168991C9810830 @default.
- W2146168991 hasLocation W21461689911 @default.
- W2146168991 hasLocation W21461689912 @default.
- W2146168991 hasLocation W21461689913 @default.
- W2146168991 hasLocation W21461689914 @default.
- W2146168991 hasLocation W21461689915 @default.
- W2146168991 hasOpenAccess W2146168991 @default.
- W2146168991 hasPrimaryLocation W21461689911 @default.
- W2146168991 hasRelatedWork W1995002252 @default.
- W2146168991 hasRelatedWork W1998825731 @default.