Matches in SemOpenAlex for { <https://semopenalex.org/work/W4221152017> ?p ?o ?g. }
Showing items 1 to 91 of
91
with 100 items per page.
- W4221152017 abstract "Compared to hybrid automatic speech recognition (ASR) systems that use a modular architecture in which each component can be independently adapted to a new domain, recent end-to-end (E2E) ASR system are harder to customize due to their all-neural monolithic construction. In this paper, we propose a novel text representation and training framework for E2E ASR models. With this approach, we show that a trained RNN Transducer (RNN-T) model's internal LM component can be effectively adapted with text-only data. An RNN-T model trained using both speech and text inputs improves over a baseline model trained on just speech with close to 13% word error rate (WER) reduction on the Switchboard and CallHome test sets of the NIST Hub5 2000 evaluation. The usefulness of the proposed approach is further demonstrated by customizing this general purpose RNN-T model to three separate datasets. We observe 20-45% relative word error rate (WER) reduction in these settings with this novel LM style customization technique using only unpaired text data from the new domains." @default.
- W4221152017 created "2022-04-03" @default.
- W4221152017 creator A5003725957 @default.
- W4221152017 creator A5056869363 @default.
- W4221152017 creator A5072442182 @default.
- W4221152017 creator A5079994647 @default.
- W4221152017 date "2022-02-26" @default.
- W4221152017 modified "2023-09-26" @default.
- W4221152017 title "Integrating Text Inputs For Training and Adapting RNN Transducer ASR Models" @default.
- W4221152017 doi "https://doi.org/10.48550/arxiv.2202.13155" @default.
- W4221152017 hasPublicationYear "2022" @default.
- W4221152017 type Work @default.
- W4221152017 citedByCount "0" @default.
- W4221152017 crossrefType "posted-content" @default.
- W4221152017 hasAuthorship W4221152017A5003725957 @default.
- W4221152017 hasAuthorship W4221152017A5056869363 @default.
- W4221152017 hasAuthorship W4221152017A5072442182 @default.
- W4221152017 hasAuthorship W4221152017A5079994647 @default.
- W4221152017 hasBestOaLocation W42211520171 @default.
- W4221152017 hasConcept C101468663 @default.
- W4221152017 hasConcept C111219384 @default.
- W4221152017 hasConcept C111335779 @default.
- W4221152017 hasConcept C111919701 @default.
- W4221152017 hasConcept C121332964 @default.
- W4221152017 hasConcept C136764020 @default.
- W4221152017 hasConcept C137293760 @default.
- W4221152017 hasConcept C138885662 @default.
- W4221152017 hasConcept C147168706 @default.
- W4221152017 hasConcept C154945302 @default.
- W4221152017 hasConcept C168167062 @default.
- W4221152017 hasConcept C17744445 @default.
- W4221152017 hasConcept C183003079 @default.
- W4221152017 hasConcept C199539241 @default.
- W4221152017 hasConcept C204321447 @default.
- W4221152017 hasConcept C2524010 @default.
- W4221152017 hasConcept C2776359362 @default.
- W4221152017 hasConcept C2780416260 @default.
- W4221152017 hasConcept C28490314 @default.
- W4221152017 hasConcept C33923547 @default.
- W4221152017 hasConcept C40969351 @default.
- W4221152017 hasConcept C41008148 @default.
- W4221152017 hasConcept C41895202 @default.
- W4221152017 hasConcept C50644808 @default.
- W4221152017 hasConcept C77088390 @default.
- W4221152017 hasConcept C90805587 @default.
- W4221152017 hasConcept C94625758 @default.
- W4221152017 hasConcept C97355855 @default.
- W4221152017 hasConceptScore W4221152017C101468663 @default.
- W4221152017 hasConceptScore W4221152017C111219384 @default.
- W4221152017 hasConceptScore W4221152017C111335779 @default.
- W4221152017 hasConceptScore W4221152017C111919701 @default.
- W4221152017 hasConceptScore W4221152017C121332964 @default.
- W4221152017 hasConceptScore W4221152017C136764020 @default.
- W4221152017 hasConceptScore W4221152017C137293760 @default.
- W4221152017 hasConceptScore W4221152017C138885662 @default.
- W4221152017 hasConceptScore W4221152017C147168706 @default.
- W4221152017 hasConceptScore W4221152017C154945302 @default.
- W4221152017 hasConceptScore W4221152017C168167062 @default.
- W4221152017 hasConceptScore W4221152017C17744445 @default.
- W4221152017 hasConceptScore W4221152017C183003079 @default.
- W4221152017 hasConceptScore W4221152017C199539241 @default.
- W4221152017 hasConceptScore W4221152017C204321447 @default.
- W4221152017 hasConceptScore W4221152017C2524010 @default.
- W4221152017 hasConceptScore W4221152017C2776359362 @default.
- W4221152017 hasConceptScore W4221152017C2780416260 @default.
- W4221152017 hasConceptScore W4221152017C28490314 @default.
- W4221152017 hasConceptScore W4221152017C33923547 @default.
- W4221152017 hasConceptScore W4221152017C40969351 @default.
- W4221152017 hasConceptScore W4221152017C41008148 @default.
- W4221152017 hasConceptScore W4221152017C41895202 @default.
- W4221152017 hasConceptScore W4221152017C50644808 @default.
- W4221152017 hasConceptScore W4221152017C77088390 @default.
- W4221152017 hasConceptScore W4221152017C90805587 @default.
- W4221152017 hasConceptScore W4221152017C94625758 @default.
- W4221152017 hasConceptScore W4221152017C97355855 @default.
- W4221152017 hasLocation W42211520171 @default.
- W4221152017 hasOpenAccess W4221152017 @default.
- W4221152017 hasPrimaryLocation W42211520171 @default.
- W4221152017 hasRelatedWork W179875071 @default.
- W4221152017 hasRelatedWork W2003136674 @default.
- W4221152017 hasRelatedWork W2244609359 @default.
- W4221152017 hasRelatedWork W2963414781 @default.
- W4221152017 hasRelatedWork W3136989387 @default.
- W4221152017 hasRelatedWork W3163300396 @default.
- W4221152017 hasRelatedWork W4221152017 @default.
- W4221152017 hasRelatedWork W4225308107 @default.
- W4221152017 hasRelatedWork W4287266619 @default.
- W4221152017 hasRelatedWork W2116192687 @default.
- W4221152017 isParatext "false" @default.
- W4221152017 isRetracted "false" @default.
- W4221152017 workType "article" @default.