Matches in SemOpenAlex for { <https://semopenalex.org/work/W4289446129> ?p ?o ?g. }
Showing items 1 to 73 of
73
with 100 items per page.
- W4289446129 abstract "This paper proposes a new approach to perform unsupervised fine-tuning and self-training using unlabeled speech data for recurrent neural network (RNN)-Transducer (RNN-T) end-to-end (E2E) automatic speech recognition (ASR) systems. Conventional systems perform fine-tuning/self-training using ASR hypothesis as the targets when using unlabeled audio data and are susceptible to the ASR performance of the base model. Here in order to alleviate the influence of ASR errors while using unlabeled data, we propose a multiple-hypothesis RNN-T loss that incorporates multiple ASR 1-best hypotheses into the loss function. For the fine-tuning task, ASR experiments on Librispeech show that the multiple-hypothesis approach achieves a relative reduction of 14.2% word error rate (WER) when compared to the single-hypothesis approach, on the test_other set. For the self-training task, ASR models are trained using supervised data from Wall Street Journal (WSJ), Aurora-4 along with CHiME-4 real noisy data as unlabeled data. The multiple-hypothesis approach yields a relative reduction of 3.3% WER on the CHiME-4's single-channel real noisy evaluation set when compared with the single-hypothesis approach." @default.
- W4289446129 created "2022-08-02" @default.
- W4289446129 creator A5057188300 @default.
- W4289446129 creator A5066876667 @default.
- W4289446129 creator A5086312107 @default.
- W4289446129 date "2022-07-29" @default.
- W4289446129 modified "2023-09-24" @default.
- W4289446129 title "Multiple-hypothesis RNN-T Loss for Unsupervised Fine-tuning and Self-training of Neural Transducer" @default.
- W4289446129 doi "https://doi.org/10.48550/arxiv.2207.14736" @default.
- W4289446129 hasPublicationYear "2022" @default.
- W4289446129 type Work @default.
- W4289446129 citedByCount "0" @default.
- W4289446129 crossrefType "posted-content" @default.
- W4289446129 hasAuthorship W4289446129A5057188300 @default.
- W4289446129 hasAuthorship W4289446129A5066876667 @default.
- W4289446129 hasAuthorship W4289446129A5086312107 @default.
- W4289446129 hasBestOaLocation W42894461291 @default.
- W4289446129 hasConcept C111335779 @default.
- W4289446129 hasConcept C147168706 @default.
- W4289446129 hasConcept C153180895 @default.
- W4289446129 hasConcept C154945302 @default.
- W4289446129 hasConcept C162324750 @default.
- W4289446129 hasConcept C16910744 @default.
- W4289446129 hasConcept C169903167 @default.
- W4289446129 hasConcept C177264268 @default.
- W4289446129 hasConcept C187736073 @default.
- W4289446129 hasConcept C199360897 @default.
- W4289446129 hasConcept C2524010 @default.
- W4289446129 hasConcept C2780451532 @default.
- W4289446129 hasConcept C28490314 @default.
- W4289446129 hasConcept C33923547 @default.
- W4289446129 hasConcept C40969351 @default.
- W4289446129 hasConcept C41008148 @default.
- W4289446129 hasConcept C50644808 @default.
- W4289446129 hasConcept C51632099 @default.
- W4289446129 hasConcept C58489278 @default.
- W4289446129 hasConcept C90805587 @default.
- W4289446129 hasConceptScore W4289446129C111335779 @default.
- W4289446129 hasConceptScore W4289446129C147168706 @default.
- W4289446129 hasConceptScore W4289446129C153180895 @default.
- W4289446129 hasConceptScore W4289446129C154945302 @default.
- W4289446129 hasConceptScore W4289446129C162324750 @default.
- W4289446129 hasConceptScore W4289446129C16910744 @default.
- W4289446129 hasConceptScore W4289446129C169903167 @default.
- W4289446129 hasConceptScore W4289446129C177264268 @default.
- W4289446129 hasConceptScore W4289446129C187736073 @default.
- W4289446129 hasConceptScore W4289446129C199360897 @default.
- W4289446129 hasConceptScore W4289446129C2524010 @default.
- W4289446129 hasConceptScore W4289446129C2780451532 @default.
- W4289446129 hasConceptScore W4289446129C28490314 @default.
- W4289446129 hasConceptScore W4289446129C33923547 @default.
- W4289446129 hasConceptScore W4289446129C40969351 @default.
- W4289446129 hasConceptScore W4289446129C41008148 @default.
- W4289446129 hasConceptScore W4289446129C50644808 @default.
- W4289446129 hasConceptScore W4289446129C51632099 @default.
- W4289446129 hasConceptScore W4289446129C58489278 @default.
- W4289446129 hasConceptScore W4289446129C90805587 @default.
- W4289446129 hasLocation W42894461291 @default.
- W4289446129 hasOpenAccess W4289446129 @default.
- W4289446129 hasPrimaryLocation W42894461291 @default.
- W4289446129 hasRelatedWork W10719664 @default.
- W4289446129 hasRelatedWork W11584726 @default.
- W4289446129 hasRelatedWork W14587445 @default.
- W4289446129 hasRelatedWork W2883085 @default.
- W4289446129 hasRelatedWork W4771408 @default.
- W4289446129 hasRelatedWork W5769393 @default.
- W4289446129 hasRelatedWork W5939791 @default.
- W4289446129 hasRelatedWork W7958345 @default.
- W4289446129 hasRelatedWork W7982726 @default.
- W4289446129 hasRelatedWork W8738421 @default.
- W4289446129 isParatext "false" @default.
- W4289446129 isRetracted "false" @default.
- W4289446129 workType "article" @default.