Matches in SemOpenAlex for { <https://semopenalex.org/work/W2892009249> ?p ?o ?g. }
Showing items 1 to 86 of
86
with 100 items per page.
- W2892009249 abstract "Recurrent sequence-to-sequence models using encoder-decoder architecture have made great progress in speech recognition task. However, they suffer from the drawback of slow training speed because the internal recurrence limits the training parallelization. In this paper, we present the Speech-Transformer, a no-recurrence sequence-to-sequence model entirely relies on attention mechanisms to learn the positional dependencies, which can be trained faster with more efficiency. We also propose a 2D-Attention mechanism, which can jointly attend to the time and frequency axes of the 2-dimensional speech inputs, thus providing more expressive representations for the Speech-Transformer. Evaluated on the Wall Street Journal (WSJ) speech recognition dataset, our best model achieves competitive word error rate (WER) of 10.9%, while the whole training process only takes 1.2 days on 1 GPU, significantly faster than the published results of recurrent sequence-to-sequence models." @default.
- W2892009249 created "2018-09-27" @default.
- W2892009249 creator A5000236213 @default.
- W2892009249 creator A5004363895 @default.
- W2892009249 creator A5068469832 @default.
- W2892009249 date "2018-04-01" @default.
- W2892009249 modified "2023-10-16" @default.
- W2892009249 title "Speech-Transformer: A No-Recurrence Sequence-to-Sequence Model for Speech Recognition" @default.
- W2892009249 cites W2125838338 @default.
- W2892009249 cites W2143612262 @default.
- W2892009249 cites W2327501763 @default.
- W2892009249 cites W2396144419 @default.
- W2892009249 cites W2526425061 @default.
- W2892009249 cites W2530876040 @default.
- W2892009249 cites W2735006420 @default.
- W2892009249 cites W2962826786 @default.
- W2892009249 doi "https://doi.org/10.1109/icassp.2018.8462506" @default.
- W2892009249 hasPublicationYear "2018" @default.
- W2892009249 type Work @default.
- W2892009249 sameAs 2892009249 @default.
- W2892009249 citedByCount "543" @default.
- W2892009249 countsByYear W28920092492018 @default.
- W2892009249 countsByYear W28920092492019 @default.
- W2892009249 countsByYear W28920092492020 @default.
- W2892009249 countsByYear W28920092492021 @default.
- W2892009249 countsByYear W28920092492022 @default.
- W2892009249 countsByYear W28920092492023 @default.
- W2892009249 crossrefType "proceedings-article" @default.
- W2892009249 hasAuthorship W2892009249A5000236213 @default.
- W2892009249 hasAuthorship W2892009249A5004363895 @default.
- W2892009249 hasAuthorship W2892009249A5068469832 @default.
- W2892009249 hasConcept C111919701 @default.
- W2892009249 hasConcept C11413529 @default.
- W2892009249 hasConcept C118505674 @default.
- W2892009249 hasConcept C119599485 @default.
- W2892009249 hasConcept C127413603 @default.
- W2892009249 hasConcept C154945302 @default.
- W2892009249 hasConcept C165801399 @default.
- W2892009249 hasConcept C201995342 @default.
- W2892009249 hasConcept C204321447 @default.
- W2892009249 hasConcept C2778112365 @default.
- W2892009249 hasConcept C2780451532 @default.
- W2892009249 hasConcept C28490314 @default.
- W2892009249 hasConcept C35639132 @default.
- W2892009249 hasConcept C40969351 @default.
- W2892009249 hasConcept C41008148 @default.
- W2892009249 hasConcept C54355233 @default.
- W2892009249 hasConcept C57273362 @default.
- W2892009249 hasConcept C66322947 @default.
- W2892009249 hasConcept C86803240 @default.
- W2892009249 hasConceptScore W2892009249C111919701 @default.
- W2892009249 hasConceptScore W2892009249C11413529 @default.
- W2892009249 hasConceptScore W2892009249C118505674 @default.
- W2892009249 hasConceptScore W2892009249C119599485 @default.
- W2892009249 hasConceptScore W2892009249C127413603 @default.
- W2892009249 hasConceptScore W2892009249C154945302 @default.
- W2892009249 hasConceptScore W2892009249C165801399 @default.
- W2892009249 hasConceptScore W2892009249C201995342 @default.
- W2892009249 hasConceptScore W2892009249C204321447 @default.
- W2892009249 hasConceptScore W2892009249C2778112365 @default.
- W2892009249 hasConceptScore W2892009249C2780451532 @default.
- W2892009249 hasConceptScore W2892009249C28490314 @default.
- W2892009249 hasConceptScore W2892009249C35639132 @default.
- W2892009249 hasConceptScore W2892009249C40969351 @default.
- W2892009249 hasConceptScore W2892009249C41008148 @default.
- W2892009249 hasConceptScore W2892009249C54355233 @default.
- W2892009249 hasConceptScore W2892009249C57273362 @default.
- W2892009249 hasConceptScore W2892009249C66322947 @default.
- W2892009249 hasConceptScore W2892009249C86803240 @default.
- W2892009249 hasLocation W28920092491 @default.
- W2892009249 hasOpenAccess W2892009249 @default.
- W2892009249 hasPrimaryLocation W28920092491 @default.
- W2892009249 hasRelatedWork W2081647779 @default.
- W2892009249 hasRelatedWork W2100336732 @default.
- W2892009249 hasRelatedWork W2892009249 @default.
- W2892009249 hasRelatedWork W2937148911 @default.
- W2892009249 hasRelatedWork W2963850025 @default.
- W2892009249 hasRelatedWork W3012645992 @default.
- W2892009249 hasRelatedWork W3015671919 @default.
- W2892009249 hasRelatedWork W3080496452 @default.
- W2892009249 hasRelatedWork W3156915121 @default.
- W2892009249 hasRelatedWork W4286421016 @default.
- W2892009249 isParatext "false" @default.
- W2892009249 isRetracted "false" @default.
- W2892009249 magId "2892009249" @default.
- W2892009249 workType "article" @default.