Matches in SemOpenAlex for { <https://semopenalex.org/work/W3149509723> ?p ?o ?g. }
- W3149509723 abstract "There is a growing interest in the speech community in developing Recurrent Neural Network Transducer (RNN-T) models for automatic speech recognition (ASR) applications. RNN-T is trained with a loss function that does not enforce temporal alignment of the training transcripts and audio. As a result, RNN-T models built with uni-directional long short term memory (LSTM) encoders tend to wait for longer spans of input audio, before streaming already decoded ASR tokens. In this work, we propose a modification to the RNN-T loss function and develop Alignment Restricted RNN-T (Ar-RNN-T) models, which utilize audio-text alignment in-formation to guide the loss computation. We compare the proposed method with existing works, such as monotonic RNN-T, on LibriSpeech and in-house datasets. We show that the Ar-RNN-T loss provides a refined control to navigate the trade-offs between the token emission delays and the Word Error Rate (WER). The Ar-RNN-T models also improve downstream applications such as the ASR End-pointing by guaranteeing token emissions within any given range of latency. Moreover, the Ar-RNN-T loss allows for bigger batch sizes and 4 times higher throughput for our LSTM model architecture, enabling faster training and convergence on GPUs." @default.
- W3149509723 created "2021-04-13" @default.
- W3149509723 creator A5015663174 @default.
- W3149509723 creator A5041313589 @default.
- W3149509723 creator A5047073253 @default.
- W3149509723 creator A5047358828 @default.
- W3149509723 creator A5048538280 @default.
- W3149509723 creator A5058126062 @default.
- W3149509723 creator A5062378252 @default.
- W3149509723 creator A5074237839 @default.
- W3149509723 creator A5087421101 @default.
- W3149509723 date "2021-01-19" @default.
- W3149509723 modified "2023-10-07" @default.
- W3149509723 title "Alignment Restricted Streaming Recurrent Neural Network Transducer" @default.
- W3149509723 cites W1494198834 @default.
- W3149509723 cites W2127141656 @default.
- W3149509723 cites W2143612262 @default.
- W3149509723 cites W2514741789 @default.
- W3149509723 cites W2625979394 @default.
- W3149509723 cites W2746192915 @default.
- W3149509723 cites W2766219058 @default.
- W3149509723 cites W2933138175 @default.
- W3149509723 cites W2935756939 @default.
- W3149509723 cites W2936774411 @default.
- W3149509723 cites W2962760690 @default.
- W3149509723 cites W2963250244 @default.
- W3149509723 cites W2963382687 @default.
- W3149509723 cites W2963414781 @default.
- W3149509723 cites W2964084166 @default.
- W3149509723 cites W3007227084 @default.
- W3149509723 cites W3007528493 @default.
- W3149509723 cites W3008174054 @default.
- W3149509723 cites W3008525923 @default.
- W3149509723 cites W3008898571 @default.
- W3149509723 cites W3015194534 @default.
- W3149509723 cites W3015315932 @default.
- W3149509723 cites W3015927303 @default.
- W3149509723 cites W3016234571 @default.
- W3149509723 cites W3094667432 @default.
- W3149509723 cites W3096888553 @default.
- W3149509723 doi "https://doi.org/10.1109/slt48900.2021.9383606" @default.
- W3149509723 hasPublicationYear "2021" @default.
- W3149509723 type Work @default.
- W3149509723 sameAs 3149509723 @default.
- W3149509723 citedByCount "34" @default.
- W3149509723 countsByYear W31495097232020 @default.
- W3149509723 countsByYear W31495097232021 @default.
- W3149509723 countsByYear W31495097232022 @default.
- W3149509723 countsByYear W31495097232023 @default.
- W3149509723 crossrefType "proceedings-article" @default.
- W3149509723 hasAuthorship W3149509723A5015663174 @default.
- W3149509723 hasAuthorship W3149509723A5041313589 @default.
- W3149509723 hasAuthorship W3149509723A5047073253 @default.
- W3149509723 hasAuthorship W3149509723A5047358828 @default.
- W3149509723 hasAuthorship W3149509723A5048538280 @default.
- W3149509723 hasAuthorship W3149509723A5058126062 @default.
- W3149509723 hasAuthorship W3149509723A5062378252 @default.
- W3149509723 hasAuthorship W3149509723A5074237839 @default.
- W3149509723 hasAuthorship W3149509723A5087421101 @default.
- W3149509723 hasBestOaLocation W31495097232 @default.
- W3149509723 hasConcept C111919701 @default.
- W3149509723 hasConcept C118505674 @default.
- W3149509723 hasConcept C137293760 @default.
- W3149509723 hasConcept C147168706 @default.
- W3149509723 hasConcept C154945302 @default.
- W3149509723 hasConcept C28490314 @default.
- W3149509723 hasConcept C31258907 @default.
- W3149509723 hasConcept C41008148 @default.
- W3149509723 hasConcept C46637626 @default.
- W3149509723 hasConcept C48145219 @default.
- W3149509723 hasConcept C50644808 @default.
- W3149509723 hasConcept C76155785 @default.
- W3149509723 hasConcept C82876162 @default.
- W3149509723 hasConceptScore W3149509723C111919701 @default.
- W3149509723 hasConceptScore W3149509723C118505674 @default.
- W3149509723 hasConceptScore W3149509723C137293760 @default.
- W3149509723 hasConceptScore W3149509723C147168706 @default.
- W3149509723 hasConceptScore W3149509723C154945302 @default.
- W3149509723 hasConceptScore W3149509723C28490314 @default.
- W3149509723 hasConceptScore W3149509723C31258907 @default.
- W3149509723 hasConceptScore W3149509723C41008148 @default.
- W3149509723 hasConceptScore W3149509723C46637626 @default.
- W3149509723 hasConceptScore W3149509723C48145219 @default.
- W3149509723 hasConceptScore W3149509723C50644808 @default.
- W3149509723 hasConceptScore W3149509723C76155785 @default.
- W3149509723 hasConceptScore W3149509723C82876162 @default.
- W3149509723 hasLocation W31495097231 @default.
- W3149509723 hasLocation W31495097232 @default.
- W3149509723 hasOpenAccess W3149509723 @default.
- W3149509723 hasPrimaryLocation W31495097231 @default.
- W3149509723 hasRelatedWork W2547835662 @default.
- W3149509723 hasRelatedWork W2608712415 @default.
- W3149509723 hasRelatedWork W2625315266 @default.
- W3149509723 hasRelatedWork W2782005958 @default.
- W3149509723 hasRelatedWork W3089122997 @default.
- W3149509723 hasRelatedWork W3109671931 @default.
- W3149509723 hasRelatedWork W4213396958 @default.
- W3149509723 hasRelatedWork W4308166499 @default.
- W3149509723 hasRelatedWork W4320560854 @default.
- W3149509723 hasRelatedWork W4375869292 @default.