Matches in SemOpenAlex for { <https://semopenalex.org/work/W2785600113> ?p ?o ?g. }
- W2785600113 abstract "When an automatic speech recognition (ASR) system is deployed for real-world applications, it often receives only one utterance at a time for decoding. This single utterance could be of short duration depending on the ASR task. In these cases, robust estimation of speaker normalizing methods like feature-space maximum likelihood linear regression (FMLLR) and i -vectors may not be feasible. In this paper, we propose two unsupervised speaker normalization techniques—one at feature level and other at model level of acoustic modeling—to overcome the drawbacks of FMLLR and i -vectors in real-time scenarios. At feature level, we propose the use of deep neural networks (DNN) to generate pseudo-FMLLR features from time-synchronous pair of filterbank and FMLLR features. These pseudo-FMLLR features can then be used for DNN acoustic model training and decoding. At model level, we propose a generalized distillation framework, where a teacher DNN trained on FMLLR features guides the training and optimization of a student DNN trained on filterbank features. In both the proposed methods, the ambiguity in choosing the speaker-specific FMLLR transform can be reduced by augmenting i -vectors to the input filterbank features. Experiments conducted on 33-h and 110-h subsets of Switchboard corpus show that the proposed methods provide significant gains over DNNs trained on FMLLR, i -vector appended FMLLR, filterbank and i -vector appended filterbank features, in real-time scenario." @default.
- W2785600113 created "2018-02-23" @default.
- W2785600113 creator A5022349534 @default.
- W2785600113 creator A5074777875 @default.
- W2785600113 creator A5085660348 @default.
- W2785600113 date "2018-04-01" @default.
- W2785600113 modified "2023-09-26" @default.
- W2785600113 title "FMLLR Speaker Normalization With i-Vector: In Pseudo-FMLLR and Distillation Framework" @default.
- W2785600113 cites W1524333225 @default.
- W2785600113 cites W1537275613 @default.
- W2785600113 cites W1821462560 @default.
- W2785600113 cites W1973681148 @default.
- W2785600113 cites W1985368711 @default.
- W2785600113 cites W1985371235 @default.
- W2785600113 cites W1989549063 @default.
- W2785600113 cites W1993409002 @default.
- W2785600113 cites W2010362084 @default.
- W2785600113 cites W2044893557 @default.
- W2785600113 cites W2078528584 @default.
- W2785600113 cites W2079623482 @default.
- W2785600113 cites W2087006792 @default.
- W2785600113 cites W2090320273 @default.
- W2785600113 cites W2160306971 @default.
- W2785600113 cites W2173379916 @default.
- W2785600113 cites W2295119550 @default.
- W2785600113 cites W2408713104 @default.
- W2785600113 cites W2512865187 @default.
- W2785600113 cites W2515525878 @default.
- W2785600113 cites W2624390389 @default.
- W2785600113 cites W2750536063 @default.
- W2785600113 cites W2962946266 @default.
- W2785600113 cites W3023820860 @default.
- W2785600113 doi "https://doi.org/10.1109/taslp.2018.2795754" @default.
- W2785600113 hasPublicationYear "2018" @default.
- W2785600113 type Work @default.
- W2785600113 sameAs 2785600113 @default.
- W2785600113 citedByCount "4" @default.
- W2785600113 countsByYear W27856001132018 @default.
- W2785600113 countsByYear W27856001132019 @default.
- W2785600113 countsByYear W27856001132020 @default.
- W2785600113 crossrefType "journal-article" @default.
- W2785600113 hasAuthorship W2785600113A5022349534 @default.
- W2785600113 hasAuthorship W2785600113A5074777875 @default.
- W2785600113 hasAuthorship W2785600113A5085660348 @default.
- W2785600113 hasConcept C100515483 @default.
- W2785600113 hasConcept C106131492 @default.
- W2785600113 hasConcept C11413529 @default.
- W2785600113 hasConcept C133892786 @default.
- W2785600113 hasConcept C136886441 @default.
- W2785600113 hasConcept C138885662 @default.
- W2785600113 hasConcept C144024400 @default.
- W2785600113 hasConcept C153180895 @default.
- W2785600113 hasConcept C154945302 @default.
- W2785600113 hasConcept C19165224 @default.
- W2785600113 hasConcept C2776401178 @default.
- W2785600113 hasConcept C28490314 @default.
- W2785600113 hasConcept C31972630 @default.
- W2785600113 hasConcept C41008148 @default.
- W2785600113 hasConcept C41895202 @default.
- W2785600113 hasConcept C50644808 @default.
- W2785600113 hasConcept C52622490 @default.
- W2785600113 hasConcept C57273362 @default.
- W2785600113 hasConcept C83665646 @default.
- W2785600113 hasConceptScore W2785600113C100515483 @default.
- W2785600113 hasConceptScore W2785600113C106131492 @default.
- W2785600113 hasConceptScore W2785600113C11413529 @default.
- W2785600113 hasConceptScore W2785600113C133892786 @default.
- W2785600113 hasConceptScore W2785600113C136886441 @default.
- W2785600113 hasConceptScore W2785600113C138885662 @default.
- W2785600113 hasConceptScore W2785600113C144024400 @default.
- W2785600113 hasConceptScore W2785600113C153180895 @default.
- W2785600113 hasConceptScore W2785600113C154945302 @default.
- W2785600113 hasConceptScore W2785600113C19165224 @default.
- W2785600113 hasConceptScore W2785600113C2776401178 @default.
- W2785600113 hasConceptScore W2785600113C28490314 @default.
- W2785600113 hasConceptScore W2785600113C31972630 @default.
- W2785600113 hasConceptScore W2785600113C41008148 @default.
- W2785600113 hasConceptScore W2785600113C41895202 @default.
- W2785600113 hasConceptScore W2785600113C50644808 @default.
- W2785600113 hasConceptScore W2785600113C52622490 @default.
- W2785600113 hasConceptScore W2785600113C57273362 @default.
- W2785600113 hasConceptScore W2785600113C83665646 @default.
- W2785600113 hasLocation W27856001131 @default.
- W2785600113 hasOpenAccess W2785600113 @default.
- W2785600113 hasPrimaryLocation W27856001131 @default.
- W2785600113 hasRelatedWork W1496290765 @default.
- W2785600113 hasRelatedWork W1592672653 @default.
- W2785600113 hasRelatedWork W1985371235 @default.
- W2785600113 hasRelatedWork W2015633636 @default.
- W2785600113 hasRelatedWork W2053067333 @default.
- W2785600113 hasRelatedWork W2063689849 @default.
- W2785600113 hasRelatedWork W2079623482 @default.
- W2785600113 hasRelatedWork W2291725357 @default.
- W2785600113 hasRelatedWork W2296483664 @default.
- W2785600113 hasRelatedWork W2395043307 @default.
- W2785600113 hasRelatedWork W2396230943 @default.
- W2785600113 hasRelatedWork W2400957076 @default.
- W2785600113 hasRelatedWork W2605514952 @default.
- W2785600113 hasRelatedWork W2608064704 @default.
- W2785600113 hasRelatedWork W2912331391 @default.