Matches in SemOpenAlex for { <https://semopenalex.org/work/W3155427814> ?p ?o ?g. }
- W3155427814 endingPage "792" @default.
- W3155427814 startingPage "788" @default.
- W3155427814 abstract "End-to-end models have achieved impressive results on the task of automatic speech recognition (ASR). For low-resource ASR tasks, however, labeled data can hardly satisfy the demand of end-to-end models. Self-supervised acoustic pre-training has already shown its amazing ASR performance, while the transcription is still inadequate for language modeling in end-to-end models. In this work, we fuse a pre-trained acoustic encoder (wav2vec2.0) and a pre-trained linguistic encoder (BERT) into an end-to-end ASR model. The fused model only needs to learn the transfer from speech to language during fine-tuning on limited labeled data. The length of the two modalities is matched by a monotonic attention mechanism without additional parameters. Besides, a fully connected layer is introduced for the hidden mapping between modalities. We further propose a scheduled fine-tuning strategy to preserve and utilize the text context modeling ability of the pre-trained linguistic encoder. Experiments show our effective utilizing of pre-trained modules. Our model achieves better recognition performance on CALLHOME corpus (15 hours) than other end-to-end models." @default.
- W3155427814 created "2021-04-26" @default.
- W3155427814 creator A5000499123 @default.
- W3155427814 creator A5068469832 @default.
- W3155427814 creator A5077184567 @default.
- W3155427814 date "2021-01-01" @default.
- W3155427814 modified "2023-10-14" @default.
- W3155427814 title "Efficiently Fusing Pretrained Acoustic and Linguistic Encoders for Low-Resource Speech Recognition" @default.
- W3155427814 cites W1526236009 @default.
- W3155427814 cites W2127141656 @default.
- W3155427814 cites W2526425061 @default.
- W3155427814 cites W2750545698 @default.
- W3155427814 cites W2888779557 @default.
- W3155427814 cites W2889213362 @default.
- W3155427814 cites W2933138175 @default.
- W3155427814 cites W2939111082 @default.
- W3155427814 cites W2953190524 @default.
- W3155427814 cites W2962925243 @default.
- W3155427814 cites W2963362078 @default.
- W3155427814 cites W2963400424 @default.
- W3155427814 cites W2963850025 @default.
- W3155427814 cites W2972889948 @default.
- W3155427814 cites W2973180718 @default.
- W3155427814 cites W3001434439 @default.
- W3155427814 cites W3016167541 @default.
- W3155427814 cites W3095350795 @default.
- W3155427814 cites W3097882114 @default.
- W3155427814 doi "https://doi.org/10.1109/lsp.2021.3071668" @default.
- W3155427814 hasPublicationYear "2021" @default.
- W3155427814 type Work @default.
- W3155427814 sameAs 3155427814 @default.
- W3155427814 citedByCount "21" @default.
- W3155427814 countsByYear W31554278142021 @default.
- W3155427814 countsByYear W31554278142022 @default.
- W3155427814 countsByYear W31554278142023 @default.
- W3155427814 crossrefType "journal-article" @default.
- W3155427814 hasAuthorship W3155427814A5000499123 @default.
- W3155427814 hasAuthorship W3155427814A5068469832 @default.
- W3155427814 hasAuthorship W3155427814A5077184567 @default.
- W3155427814 hasBestOaLocation W31554278142 @default.
- W3155427814 hasConcept C111919701 @default.
- W3155427814 hasConcept C118505674 @default.
- W3155427814 hasConcept C137293760 @default.
- W3155427814 hasConcept C138885662 @default.
- W3155427814 hasConcept C151730666 @default.
- W3155427814 hasConcept C154945302 @default.
- W3155427814 hasConcept C155635449 @default.
- W3155427814 hasConcept C162324750 @default.
- W3155427814 hasConcept C179926584 @default.
- W3155427814 hasConcept C187736073 @default.
- W3155427814 hasConcept C204321447 @default.
- W3155427814 hasConcept C2776145971 @default.
- W3155427814 hasConcept C2779343474 @default.
- W3155427814 hasConcept C2780451532 @default.
- W3155427814 hasConcept C28490314 @default.
- W3155427814 hasConcept C41008148 @default.
- W3155427814 hasConcept C41895202 @default.
- W3155427814 hasConcept C61328038 @default.
- W3155427814 hasConcept C74296488 @default.
- W3155427814 hasConcept C86803240 @default.
- W3155427814 hasConceptScore W3155427814C111919701 @default.
- W3155427814 hasConceptScore W3155427814C118505674 @default.
- W3155427814 hasConceptScore W3155427814C137293760 @default.
- W3155427814 hasConceptScore W3155427814C138885662 @default.
- W3155427814 hasConceptScore W3155427814C151730666 @default.
- W3155427814 hasConceptScore W3155427814C154945302 @default.
- W3155427814 hasConceptScore W3155427814C155635449 @default.
- W3155427814 hasConceptScore W3155427814C162324750 @default.
- W3155427814 hasConceptScore W3155427814C179926584 @default.
- W3155427814 hasConceptScore W3155427814C187736073 @default.
- W3155427814 hasConceptScore W3155427814C204321447 @default.
- W3155427814 hasConceptScore W3155427814C2776145971 @default.
- W3155427814 hasConceptScore W3155427814C2779343474 @default.
- W3155427814 hasConceptScore W3155427814C2780451532 @default.
- W3155427814 hasConceptScore W3155427814C28490314 @default.
- W3155427814 hasConceptScore W3155427814C41008148 @default.
- W3155427814 hasConceptScore W3155427814C41895202 @default.
- W3155427814 hasConceptScore W3155427814C61328038 @default.
- W3155427814 hasConceptScore W3155427814C74296488 @default.
- W3155427814 hasConceptScore W3155427814C86803240 @default.
- W3155427814 hasFunder F4320336026 @default.
- W3155427814 hasLocation W31554278141 @default.
- W3155427814 hasLocation W31554278142 @default.
- W3155427814 hasLocation W31554278143 @default.
- W3155427814 hasOpenAccess W3155427814 @default.
- W3155427814 hasPrimaryLocation W31554278141 @default.
- W3155427814 hasRelatedWork W2126322296 @default.
- W3155427814 hasRelatedWork W2140351598 @default.
- W3155427814 hasRelatedWork W2163537793 @default.
- W3155427814 hasRelatedWork W2402899696 @default.
- W3155427814 hasRelatedWork W2781555308 @default.
- W3155427814 hasRelatedWork W2916997151 @default.
- W3155427814 hasRelatedWork W2949174760 @default.
- W3155427814 hasRelatedWork W2955724459 @default.
- W3155427814 hasRelatedWork W3021690593 @default.
- W3155427814 hasRelatedWork W3198455051 @default.
- W3155427814 hasVolume "28" @default.
- W3155427814 isParatext "false" @default.