Matches in SemOpenAlex for { <https://semopenalex.org/work/W2022608894> ?p ?o ?g. }
Showing items 1 to 97 of
97
with 100 items per page.
- W2022608894 abstract "In the last years, latent variable models such as factor analysis, probabilistic principal component analysis or subspace Gaussian mixture models have become almost ubiquitous in speech technologies. The key to its success is the joint modeling of multiple effects in the speech signal they address. In this paper, we propose a novel approach to use phone and speaker variabilities together to estimate phone posterior probabilities on a tandem speech recognition system. A Multilayer Perceptron (MLP) with 5 layers and a central bottleneck linear layer is used as a basic processing block that mimics the processing undergone in factor analysis. With multiple factors, phone and a speaker MLP are merged at the bottleneck level to obtain better estimates for the phone posterior probabilities used in the ASR system. Experiments on the WSJ corpus show that the joint phone-speaker modeling can significantly outperform phone modeling alone in terms of Frame Error and Word Error Rates." @default.
- W2022608894 created "2016-06-24" @default.
- W2022608894 creator A5027284390 @default.
- W2022608894 creator A5047712456 @default.
- W2022608894 date "2013-05-01" @default.
- W2022608894 modified "2023-09-26" @default.
- W2022608894 title "MLP-based factor analysis for tandem speech recognition" @default.
- W2022608894 cites W111477576 @default.
- W2022608894 cites W1484387534 @default.
- W2022608894 cites W1605997703 @default.
- W2022608894 cites W1846206282 @default.
- W2022608894 cites W1981706894 @default.
- W2022608894 cites W2012897754 @default.
- W2022608894 cites W2107638917 @default.
- W2022608894 cites W2121415728 @default.
- W2022608894 cites W2121812409 @default.
- W2022608894 cites W2122538988 @default.
- W2022608894 cites W2136879537 @default.
- W2022608894 cites W2160815625 @default.
- W2022608894 cites W2184045248 @default.
- W2022608894 doi "https://doi.org/10.1109/icassp.2013.6638962" @default.
- W2022608894 hasPublicationYear "2013" @default.
- W2022608894 type Work @default.
- W2022608894 sameAs 2022608894 @default.
- W2022608894 citedByCount "12" @default.
- W2022608894 countsByYear W20226088942013 @default.
- W2022608894 countsByYear W20226088942014 @default.
- W2022608894 countsByYear W20226088942015 @default.
- W2022608894 countsByYear W20226088942016 @default.
- W2022608894 crossrefType "proceedings-article" @default.
- W2022608894 hasAuthorship W2022608894A5027284390 @default.
- W2022608894 hasAuthorship W2022608894A5047712456 @default.
- W2022608894 hasConcept C107673813 @default.
- W2022608894 hasConcept C114289077 @default.
- W2022608894 hasConcept C138885662 @default.
- W2022608894 hasConcept C149635348 @default.
- W2022608894 hasConcept C153180895 @default.
- W2022608894 hasConcept C154945302 @default.
- W2022608894 hasConcept C179717631 @default.
- W2022608894 hasConcept C23224414 @default.
- W2022608894 hasConcept C27438332 @default.
- W2022608894 hasConcept C2778707766 @default.
- W2022608894 hasConcept C2780513914 @default.
- W2022608894 hasConcept C28490314 @default.
- W2022608894 hasConcept C40969351 @default.
- W2022608894 hasConcept C41008148 @default.
- W2022608894 hasConcept C41895202 @default.
- W2022608894 hasConcept C50644808 @default.
- W2022608894 hasConcept C57830394 @default.
- W2022608894 hasConcept C61328038 @default.
- W2022608894 hasConcept C70437156 @default.
- W2022608894 hasConceptScore W2022608894C107673813 @default.
- W2022608894 hasConceptScore W2022608894C114289077 @default.
- W2022608894 hasConceptScore W2022608894C138885662 @default.
- W2022608894 hasConceptScore W2022608894C149635348 @default.
- W2022608894 hasConceptScore W2022608894C153180895 @default.
- W2022608894 hasConceptScore W2022608894C154945302 @default.
- W2022608894 hasConceptScore W2022608894C179717631 @default.
- W2022608894 hasConceptScore W2022608894C23224414 @default.
- W2022608894 hasConceptScore W2022608894C27438332 @default.
- W2022608894 hasConceptScore W2022608894C2778707766 @default.
- W2022608894 hasConceptScore W2022608894C2780513914 @default.
- W2022608894 hasConceptScore W2022608894C28490314 @default.
- W2022608894 hasConceptScore W2022608894C40969351 @default.
- W2022608894 hasConceptScore W2022608894C41008148 @default.
- W2022608894 hasConceptScore W2022608894C41895202 @default.
- W2022608894 hasConceptScore W2022608894C50644808 @default.
- W2022608894 hasConceptScore W2022608894C57830394 @default.
- W2022608894 hasConceptScore W2022608894C61328038 @default.
- W2022608894 hasConceptScore W2022608894C70437156 @default.
- W2022608894 hasLocation W20226088941 @default.
- W2022608894 hasOpenAccess W2022608894 @default.
- W2022608894 hasPrimaryLocation W20226088941 @default.
- W2022608894 hasRelatedWork W1524333225 @default.
- W2022608894 hasRelatedWork W1550433238 @default.
- W2022608894 hasRelatedWork W1894518610 @default.
- W2022608894 hasRelatedWork W1993409002 @default.
- W2022608894 hasRelatedWork W2014345428 @default.
- W2022608894 hasRelatedWork W2042010990 @default.
- W2022608894 hasRelatedWork W2054584253 @default.
- W2022608894 hasRelatedWork W2079623482 @default.
- W2022608894 hasRelatedWork W2104804288 @default.
- W2022608894 hasRelatedWork W2109560142 @default.
- W2022608894 hasRelatedWork W2116338171 @default.
- W2022608894 hasRelatedWork W2122279754 @default.
- W2022608894 hasRelatedWork W2134519949 @default.
- W2022608894 hasRelatedWork W2147768505 @default.
- W2022608894 hasRelatedWork W2152228006 @default.
- W2022608894 hasRelatedWork W2156373089 @default.
- W2022608894 hasRelatedWork W2160306971 @default.
- W2022608894 hasRelatedWork W2160815625 @default.
- W2022608894 hasRelatedWork W2167386623 @default.
- W2022608894 hasRelatedWork W2184045248 @default.
- W2022608894 isParatext "false" @default.
- W2022608894 isRetracted "false" @default.
- W2022608894 magId "2022608894" @default.
- W2022608894 workType "article" @default.