END-TO-END UZBEK-RUSSIAN SPEECH TRANSLATION WITH SELF-SUPERVISED PRETRAINING

dc.contributor.author	Sukhrob Avezov Sobirovich
dc.date.accessioned	2025-12-29T14:28:39Z
dc.date.issued	2025-09-27
dc.description.abstract	In this article we study end-to-end Uzbek→Russian speech translation under realistic low-resource and code-switching conditions. We couple a wav2vec-style encoder pre-trained on unlabeled audio with a Transformer decoder, add multi-task ASR/CTC objectives, and distill from a strong cascade teacher. Script-aware tokenization and data augmentation reduce sparsity. On conversational and broadcast tests the model improves BLEU/chrF at fixed latency and yields fewer morphology and NE errors.
dc.format	application/pdf
dc.identifier.uri	https://webofjournals.com/index.php/1/article/view/5127
dc.identifier.uri	https://asianeducationindex.com/handle/123456789/22185
dc.language.iso	eng
dc.publisher	Web of Journals Publishing
dc.relation	https://webofjournals.com/index.php/1/article/view/5127/5162
dc.rights	https://creativecommons.org/licenses/by-nc-nd/4.0
dc.source	Web of Teachers: Inderscience Research ; Vol. 3 No. 9 (2025): WOT; 79-82
dc.source	2938-379X
dc.subject	End-to-end speech translation, Uzbek-Russian, self-supervised pretraining, wav2vec 2.0, XLS-R, knowledge distillation, code-switching, low-resource.
dc.title	END-TO-END UZBEK-RUSSIAN SPEECH TRANSLATION WITH SELF-SUPERVISED PRETRAINING
dc.type	info:eu-repo/semantics/article
dc.type	info:eu-repo/semantics/publishedVersion
dc.type	Peer-reviewed Article

item.page.files

pagination.showing.labelpagination.showing.detail

item.page.filesection.name: sobirovich_2025_end-to-end_uzbek-russian_speech_translat.pdf
item.page.filesection.size: 325.68 KB
item.page.filesection.format: Adobe Portable Document Format