Matches in SemOpenAlex for { <https://semopenalex.org/work/W4226543485> ?p ?o ?g. }
Showing items 1 to 75 of
75
with 100 items per page.
- W4226543485 abstract "Direct speech-to-speech translation (S2ST) models suffer from data scarcity issues as there exists little parallel S2ST data, compared to the amount of data available for conventional cascaded systems that consist of automatic speech recognition (ASR), machine translation (MT), and text-to-speech (TTS) synthesis. In this work, we explore self-supervised pre-training with unlabeled speech data and data augmentation to tackle this issue. We take advantage of a recently proposed speech-to-unit translation (S2UT) framework that encodes target speech into discrete representations, and transfer pre-training and efficient partial finetuning techniques that work well for speech-to-text translation (S2T) to the S2UT domain by studying both speech encoder and discrete unit decoder pre-training. Our experiments on Spanish-English translation show that self-supervised pre-training consistently improves model performance compared with multitask learning with an average 6.6-12.1 BLEU gain, and it can be further combined with data augmentation techniques that apply MT to create weakly supervised training data. Audio samples are available at: https://facebookresearch.github.io/speech_translation/enhanced_direct_s2st_units/index.html ." @default.
- W4226543485 created "2022-05-05" @default.
- W4226543485 creator A5005191803 @default.
- W4226543485 creator A5051950818 @default.
- W4226543485 creator A5058915697 @default.
- W4226543485 creator A5059951425 @default.
- W4226543485 creator A5072019513 @default.
- W4226543485 creator A5076914759 @default.
- W4226543485 creator A5084509956 @default.
- W4226543485 creator A5087491225 @default.
- W4226543485 date "2022-04-06" @default.
- W4226543485 modified "2023-09-25" @default.
- W4226543485 title "Enhanced Direct Speech-to-Speech Translation Using Self-supervised Pre-training and Data Augmentation" @default.
- W4226543485 doi "https://doi.org/10.48550/arxiv.2204.02967" @default.
- W4226543485 hasPublicationYear "2022" @default.
- W4226543485 type Work @default.
- W4226543485 citedByCount "0" @default.
- W4226543485 crossrefType "posted-content" @default.
- W4226543485 hasAuthorship W4226543485A5005191803 @default.
- W4226543485 hasAuthorship W4226543485A5051950818 @default.
- W4226543485 hasAuthorship W4226543485A5058915697 @default.
- W4226543485 hasAuthorship W4226543485A5059951425 @default.
- W4226543485 hasAuthorship W4226543485A5072019513 @default.
- W4226543485 hasAuthorship W4226543485A5076914759 @default.
- W4226543485 hasAuthorship W4226543485A5084509956 @default.
- W4226543485 hasAuthorship W4226543485A5087491225 @default.
- W4226543485 hasBestOaLocation W42265434851 @default.
- W4226543485 hasConcept C104317684 @default.
- W4226543485 hasConcept C105580179 @default.
- W4226543485 hasConcept C111919701 @default.
- W4226543485 hasConcept C118505674 @default.
- W4226543485 hasConcept C149364088 @default.
- W4226543485 hasConcept C14999030 @default.
- W4226543485 hasConcept C154945302 @default.
- W4226543485 hasConcept C185592680 @default.
- W4226543485 hasConcept C203005215 @default.
- W4226543485 hasConcept C204321447 @default.
- W4226543485 hasConcept C2776145971 @default.
- W4226543485 hasConcept C2780366754 @default.
- W4226543485 hasConcept C28490314 @default.
- W4226543485 hasConcept C41008148 @default.
- W4226543485 hasConcept C51632099 @default.
- W4226543485 hasConcept C55493867 @default.
- W4226543485 hasConceptScore W4226543485C104317684 @default.
- W4226543485 hasConceptScore W4226543485C105580179 @default.
- W4226543485 hasConceptScore W4226543485C111919701 @default.
- W4226543485 hasConceptScore W4226543485C118505674 @default.
- W4226543485 hasConceptScore W4226543485C149364088 @default.
- W4226543485 hasConceptScore W4226543485C14999030 @default.
- W4226543485 hasConceptScore W4226543485C154945302 @default.
- W4226543485 hasConceptScore W4226543485C185592680 @default.
- W4226543485 hasConceptScore W4226543485C203005215 @default.
- W4226543485 hasConceptScore W4226543485C204321447 @default.
- W4226543485 hasConceptScore W4226543485C2776145971 @default.
- W4226543485 hasConceptScore W4226543485C2780366754 @default.
- W4226543485 hasConceptScore W4226543485C28490314 @default.
- W4226543485 hasConceptScore W4226543485C41008148 @default.
- W4226543485 hasConceptScore W4226543485C51632099 @default.
- W4226543485 hasConceptScore W4226543485C55493867 @default.
- W4226543485 hasLocation W42265434851 @default.
- W4226543485 hasOpenAccess W4226543485 @default.
- W4226543485 hasPrimaryLocation W42265434851 @default.
- W4226543485 hasRelatedWork W1592339875 @default.
- W4226543485 hasRelatedWork W2128876910 @default.
- W4226543485 hasRelatedWork W2154135679 @default.
- W4226543485 hasRelatedWork W2435130738 @default.
- W4226543485 hasRelatedWork W2587623425 @default.
- W4226543485 hasRelatedWork W2606032440 @default.
- W4226543485 hasRelatedWork W3107474891 @default.
- W4226543485 hasRelatedWork W4226543485 @default.
- W4226543485 hasRelatedWork W4283834483 @default.
- W4226543485 hasRelatedWork W4308759026 @default.
- W4226543485 isParatext "false" @default.
- W4226543485 isRetracted "false" @default.
- W4226543485 workType "article" @default.