FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3144557079> ?p ?o ?g. }

• W3144557079 abstract "End-to-end automatic speech recognition (ASR) models with a single neural network have recently demonstrated state-of-the-art results compared to conventional hybrid speech recognizers. Specifically, recurrent neural network transducer (RNN-T) has shown competitive ASR performance on various benchmarks. In this work, we examine ways in which RNN-T can achieve better ASR accuracy via performing auxiliary tasks. We propose (i) using the same auxiliary task as primary RNN-T ASR task, and (ii) performing context-dependent graphemic state prediction as in conventional hybrid modeling. In transcribing social media videos with varying training data size, we first evaluate the streaming ASR performance on three languages: Romanian, Turkish and German. We find that both proposed methods provide consistent improvements. Next, we observe that both auxiliary tasks demonstrate efficacy in learning deep transformer encoders for RNN-T criterion, thus achieving competitive results -2.0%/4.2% WER on LibriSpeech test-clean/other - as compared to prior top performing models." @default.
• W3144557079 created "2021-04-13" @default.
• W3144557079 creator A5051165898 @default.
• W3144557079 creator A5062378252 @default.
• W3144557079 creator A5062940343 @default.
• W3144557079 creator A5069296252 @default.
• W3144557079 creator A5069954850 @default.
• W3144557079 creator A5085650357 @default.
• W3144557079 date "2021-01-19" @default.
• W3144557079 modified "2023-10-16" @default.
• W3144557079 title "Improving RNN Transducer Based ASR with Auxiliary Tasks" @default.
• W3144557079 cites W1494198834 @default.
• W3144557079 cites W1506752962 @default.
• W3144557079 cites W2097117768 @default.
• W3144557079 cites W2147768505 @default.
• W3144557079 cites W2158069733 @default.
• W3144557079 cites W2183341477 @default.
• W3144557079 cites W2194775991 @default.
• W3144557079 cites W2327501763 @default.
• W3144557079 cites W2521999726 @default.
• W3144557079 cites W2620998106 @default.
• W3144557079 cites W2746192915 @default.
• W3144557079 cites W2889030467 @default.
• W3144557079 cites W2913340405 @default.
• W3144557079 cites W2933138175 @default.
• W3144557079 cites W2936774411 @default.
• W3144557079 cites W2962760690 @default.
• W3144557079 cites W2963240019 @default.
• W3144557079 cites W2963250244 @default.
• W3144557079 cites W2963303028 @default.
• W3144557079 cites W2963414781 @default.
• W3144557079 cites W2964084166 @default.
• W3144557079 cites W2972899427 @default.
• W3144557079 cites W2981857663 @default.
• W3144557079 cites W2983490653 @default.
• W3144557079 cites W3007227084 @default.
• W3144557079 cites W3008181812 @default.
• W3144557079 cites W3008525923 @default.
• W3144557079 cites W3008898571 @default.
• W3144557079 cites W3015194534 @default.
• W3144557079 cites W3015315932 @default.
• W3144557079 cites W3015654466 @default.
• W3144557079 cites W3015960524 @default.
• W3144557079 cites W3015995734 @default.
• W3144557079 cites W3016010032 @default.
• W3144557079 cites W3035216398 @default.
• W3144557079 cites W3097558625 @default.
• W3144557079 cites W3097777922 @default.
• W3144557079 cites W3150637114 @default.
• W3144557079 doi "https://doi.org/10.1109/slt48900.2021.9383548" @default.
• W3144557079 hasPublicationYear "2021" @default.
• W3144557079 type Work @default.
• W3144557079 sameAs 3144557079 @default.
• W3144557079 citedByCount "23" @default.
• W3144557079 countsByYear W31445570792020 @default.
• W3144557079 countsByYear W31445570792021 @default.
• W3144557079 countsByYear W31445570792022 @default.
• W3144557079 countsByYear W31445570792023 @default.
• W3144557079 crossrefType "proceedings-article" @default.
• W3144557079 hasAuthorship W3144557079A5051165898 @default.
• W3144557079 hasAuthorship W3144557079A5062378252 @default.
• W3144557079 hasAuthorship W3144557079A5062940343 @default.
• W3144557079 hasAuthorship W3144557079A5069296252 @default.
• W3144557079 hasAuthorship W3144557079A5069954850 @default.
• W3144557079 hasAuthorship W3144557079A5085650357 @default.
• W3144557079 hasBestOaLocation W31445570792 @default.
• W3144557079 hasConcept C108583219 @default.
• W3144557079 hasConcept C111919701 @default.
• W3144557079 hasConcept C118505674 @default.
• W3144557079 hasConcept C121332964 @default.
• W3144557079 hasConcept C137293760 @default.
• W3144557079 hasConcept C147168706 @default.
• W3144557079 hasConcept C151730666 @default.
• W3144557079 hasConcept C154945302 @default.
• W3144557079 hasConcept C162324750 @default.
• W3144557079 hasConcept C165801399 @default.
• W3144557079 hasConcept C187736073 @default.
• W3144557079 hasConcept C2779343474 @default.
• W3144557079 hasConcept C2780451532 @default.
• W3144557079 hasConcept C28490314 @default.
• W3144557079 hasConcept C41008148 @default.
• W3144557079 hasConcept C50644808 @default.
• W3144557079 hasConcept C62520636 @default.
• W3144557079 hasConcept C66322947 @default.
• W3144557079 hasConcept C86803240 @default.
• W3144557079 hasConceptScore W3144557079C108583219 @default.
• W3144557079 hasConceptScore W3144557079C111919701 @default.
• W3144557079 hasConceptScore W3144557079C118505674 @default.
• W3144557079 hasConceptScore W3144557079C121332964 @default.
• W3144557079 hasConceptScore W3144557079C137293760 @default.
• W3144557079 hasConceptScore W3144557079C147168706 @default.
• W3144557079 hasConceptScore W3144557079C151730666 @default.
• W3144557079 hasConceptScore W3144557079C154945302 @default.
• W3144557079 hasConceptScore W3144557079C162324750 @default.
• W3144557079 hasConceptScore W3144557079C165801399 @default.
• W3144557079 hasConceptScore W3144557079C187736073 @default.
• W3144557079 hasConceptScore W3144557079C2779343474 @default.
• W3144557079 hasConceptScore W3144557079C2780451532 @default.
• W3144557079 hasConceptScore W3144557079C28490314 @default.
• W3144557079 hasConceptScore W3144557079C41008148 @default.

• next