Matches in SemOpenAlex for { <https://semopenalex.org/work/W4287083409> ?p ?o ?g. }
Showing items 1 to 53 of
53
with 100 items per page.
- W4287083409 abstract "The advent of Transformer-based models has surpassed the barriers of text. When working with speech, we must face a problem: the sequence length of an audio input is not suitable for the Transformer. To bypass this problem, a usual approach is adding strided convolutional layers, to reduce the sequence length before using the Transformer. In this paper, we propose a new approach for direct Speech Translation, where thanks to an efficient Transformer we can work with a spectrogram without having to use convolutional layers before the Transformer. This allows the encoder to learn directly from the spectrogram and no information is lost. We have created an encoder-decoder model, where the encoder is an efficient Transformer -- the Longformer -- and the decoder is a traditional Transformer decoder. Our results, which are close to the ones obtained with the standard approach, show that this is a promising research direction." @default.
- W4287083409 created "2022-07-25" @default.
- W4287083409 creator A5014771820 @default.
- W4287083409 creator A5074210163 @default.
- W4287083409 creator A5077236938 @default.
- W4287083409 date "2021-07-07" @default.
- W4287083409 modified "2023-09-27" @default.
- W4287083409 title "Efficient Transformer for Direct Speech Translation" @default.
- W4287083409 doi "https://doi.org/10.48550/arxiv.2107.03069" @default.
- W4287083409 hasPublicationYear "2021" @default.
- W4287083409 type Work @default.
- W4287083409 citedByCount "0" @default.
- W4287083409 crossrefType "posted-content" @default.
- W4287083409 hasAuthorship W4287083409A5014771820 @default.
- W4287083409 hasAuthorship W4287083409A5074210163 @default.
- W4287083409 hasAuthorship W4287083409A5077236938 @default.
- W4287083409 hasBestOaLocation W42870834091 @default.
- W4287083409 hasConcept C111919701 @default.
- W4287083409 hasConcept C118505674 @default.
- W4287083409 hasConcept C119599485 @default.
- W4287083409 hasConcept C127413603 @default.
- W4287083409 hasConcept C154945302 @default.
- W4287083409 hasConcept C165801399 @default.
- W4287083409 hasConcept C28490314 @default.
- W4287083409 hasConcept C41008148 @default.
- W4287083409 hasConcept C45273575 @default.
- W4287083409 hasConcept C66322947 @default.
- W4287083409 hasConceptScore W4287083409C111919701 @default.
- W4287083409 hasConceptScore W4287083409C118505674 @default.
- W4287083409 hasConceptScore W4287083409C119599485 @default.
- W4287083409 hasConceptScore W4287083409C127413603 @default.
- W4287083409 hasConceptScore W4287083409C154945302 @default.
- W4287083409 hasConceptScore W4287083409C165801399 @default.
- W4287083409 hasConceptScore W4287083409C28490314 @default.
- W4287083409 hasConceptScore W4287083409C41008148 @default.
- W4287083409 hasConceptScore W4287083409C45273575 @default.
- W4287083409 hasConceptScore W4287083409C66322947 @default.
- W4287083409 hasLocation W42870834091 @default.
- W4287083409 hasOpenAccess W4287083409 @default.
- W4287083409 hasPrimaryLocation W42870834091 @default.
- W4287083409 hasRelatedWork W2892009249 @default.
- W4287083409 hasRelatedWork W2959758584 @default.
- W4287083409 hasRelatedWork W3161109662 @default.
- W4287083409 hasRelatedWork W4213428484 @default.
- W4287083409 hasRelatedWork W4226419783 @default.
- W4287083409 hasRelatedWork W4286588216 @default.
- W4287083409 hasRelatedWork W4308469534 @default.
- W4287083409 hasRelatedWork W4312095844 @default.
- W4287083409 hasRelatedWork W4320016073 @default.
- W4287083409 hasRelatedWork W4323520692 @default.
- W4287083409 isParatext "false" @default.
- W4287083409 isRetracted "false" @default.
- W4287083409 workType "article" @default.