Matches in SemOpenAlex for { <https://semopenalex.org/work/W4224934179> ?p ?o ?g. }
Showing items 1 to 89 of
89
with 100 items per page.
- W4224934179 abstract "Recently, pioneer work finds that self-supervised pre-training methods can improve multiple downstream speech tasks, because the model utilizes bottom layers to learn speaker-related information and top layers to encode content-related information. Since the network capacity is limited, we believe the speech recognition performance could be further improved if the model is dedicated to audio content information learning. To this end, we propose Intermediate Layer Supervision for Self-Supervised Learning (ILS-SSL), which forces the model to concentrate on content information as much as possible by adding an additional SSL loss on the intermediate layers. Experiments on LibriSpeech test-other set show that our method outperforms HuBERT significantly, which achieves a 23.5%/11.6% relative word error rate reduction in the w/o language model setting for Base/Large models. Detailed analysis shows the bottom layers of our model have a better correlation with phonetic units, which is consistent with our intuition and explains the success of our method for ASR. We will release our code and model at https://github.com/microsoft/UniSpeech." @default.
- W4224934179 created "2022-04-28" @default.
- W4224934179 creator A5013188461 @default.
- W4224934179 creator A5015824704 @default.
- W4224934179 creator A5029670581 @default.
- W4224934179 creator A5034770439 @default.
- W4224934179 creator A5045556326 @default.
- W4224934179 creator A5067921099 @default.
- W4224934179 creator A5079533447 @default.
- W4224934179 date "2022-05-23" @default.
- W4224934179 modified "2023-09-26" @default.
- W4224934179 title "Improving Self-Supervised Learning for Speech Recognition with Intermediate Layer Supervision" @default.
- W4224934179 cites W1494198834 @default.
- W4224934179 cites W2972943112 @default.
- W4224934179 cites W2973049979 @default.
- W4224934179 cites W2973157397 @default.
- W4224934179 cites W2995181338 @default.
- W4224934179 cites W3003875258 @default.
- W4224934179 cites W3015213852 @default.
- W4224934179 cites W3016011332 @default.
- W4224934179 cites W3041561163 @default.
- W4224934179 cites W3144810982 @default.
- W4224934179 cites W3197580070 @default.
- W4224934179 cites W3209059054 @default.
- W4224934179 cites W4226033575 @default.
- W4224934179 doi "https://doi.org/10.1109/icassp43922.2022.9747022" @default.
- W4224934179 hasPublicationYear "2022" @default.
- W4224934179 type Work @default.
- W4224934179 citedByCount "3" @default.
- W4224934179 countsByYear W42249341792022 @default.
- W4224934179 countsByYear W42249341792023 @default.
- W4224934179 crossrefType "proceedings-article" @default.
- W4224934179 hasAuthorship W4224934179A5013188461 @default.
- W4224934179 hasAuthorship W4224934179A5015824704 @default.
- W4224934179 hasAuthorship W4224934179A5029670581 @default.
- W4224934179 hasAuthorship W4224934179A5034770439 @default.
- W4224934179 hasAuthorship W4224934179A5045556326 @default.
- W4224934179 hasAuthorship W4224934179A5067921099 @default.
- W4224934179 hasAuthorship W4224934179A5079533447 @default.
- W4224934179 hasConcept C104317684 @default.
- W4224934179 hasConcept C111472728 @default.
- W4224934179 hasConcept C132010649 @default.
- W4224934179 hasConcept C137293760 @default.
- W4224934179 hasConcept C138885662 @default.
- W4224934179 hasConcept C154945302 @default.
- W4224934179 hasConcept C177264268 @default.
- W4224934179 hasConcept C178790620 @default.
- W4224934179 hasConcept C185592680 @default.
- W4224934179 hasConcept C199360897 @default.
- W4224934179 hasConcept C204321447 @default.
- W4224934179 hasConcept C2779227376 @default.
- W4224934179 hasConcept C28490314 @default.
- W4224934179 hasConcept C40969351 @default.
- W4224934179 hasConcept C41008148 @default.
- W4224934179 hasConcept C55493867 @default.
- W4224934179 hasConcept C66746571 @default.
- W4224934179 hasConceptScore W4224934179C104317684 @default.
- W4224934179 hasConceptScore W4224934179C111472728 @default.
- W4224934179 hasConceptScore W4224934179C132010649 @default.
- W4224934179 hasConceptScore W4224934179C137293760 @default.
- W4224934179 hasConceptScore W4224934179C138885662 @default.
- W4224934179 hasConceptScore W4224934179C154945302 @default.
- W4224934179 hasConceptScore W4224934179C177264268 @default.
- W4224934179 hasConceptScore W4224934179C178790620 @default.
- W4224934179 hasConceptScore W4224934179C185592680 @default.
- W4224934179 hasConceptScore W4224934179C199360897 @default.
- W4224934179 hasConceptScore W4224934179C204321447 @default.
- W4224934179 hasConceptScore W4224934179C2779227376 @default.
- W4224934179 hasConceptScore W4224934179C28490314 @default.
- W4224934179 hasConceptScore W4224934179C40969351 @default.
- W4224934179 hasConceptScore W4224934179C41008148 @default.
- W4224934179 hasConceptScore W4224934179C55493867 @default.
- W4224934179 hasConceptScore W4224934179C66746571 @default.
- W4224934179 hasLocation W42249341791 @default.
- W4224934179 hasOpenAccess W4224934179 @default.
- W4224934179 hasPrimaryLocation W42249341791 @default.
- W4224934179 hasRelatedWork W1546181631 @default.
- W4224934179 hasRelatedWork W1989705153 @default.
- W4224934179 hasRelatedWork W2047094413 @default.
- W4224934179 hasRelatedWork W2100441082 @default.
- W4224934179 hasRelatedWork W2166817309 @default.
- W4224934179 hasRelatedWork W3037333170 @default.
- W4224934179 hasRelatedWork W3080136773 @default.
- W4224934179 hasRelatedWork W3107474891 @default.
- W4224934179 hasRelatedWork W4224934179 @default.
- W4224934179 hasRelatedWork W4226221575 @default.
- W4224934179 isParatext "false" @default.
- W4224934179 isRetracted "false" @default.
- W4224934179 workType "article" @default.