Matches in SemOpenAlex for { <https://semopenalex.org/work/W3002961036> ?p ?o ?g. }
- W3002961036 abstract "Recently, Transformer has gained success in automatic speech recognition (ASR) field. However, it is challenging to deploy a Transformer-based end-to-end (E2E) model for online speech recognition. In this paper, we propose the Transformer-based online CTC/attention E2E ASR architecture, which contains the chunk self-attention encoder (chunk-SAE) and the monotonic truncated attention (MTA) based self-attention decoder (SAD). Firstly, the chunk-SAE splits the speech into isolated chunks. To reduce the computational cost and improve the performance, we propose the state reuse chunk-SAE. Sencondly, the MTA based SAD truncates the speech features monotonically and performs attention on the truncated features. To support the online recognition, we integrate the state reuse chunk-SAE and the MTA based SAD into online CTC/attention architecture. We evaluate the proposed online models on the HKUST Mandarin ASR benchmark and achieve a 23.66% character error rate (CER) with a 320 ms latency. Our online model yields as little as 0.19% absolute CER degradation compared with the offline baseline, and achieves significant improvement over our prior work on Long Short-Term Memory (LSTM) based online E2E models." @default.
- W3002961036 created "2020-01-30" @default.
- W3002961036 creator A5006453801 @default.
- W3002961036 creator A5027112640 @default.
- W3002961036 creator A5036446253 @default.
- W3002961036 creator A5066595006 @default.
- W3002961036 creator A5072887461 @default.
- W3002961036 date "2020-01-15" @default.
- W3002961036 modified "2023-09-25" @default.
- W3002961036 title "Transformer-based Online CTC/attention End-to-End Speech Recognition Architecture" @default.
- W3002961036 cites W1526236009 @default.
- W3002961036 cites W2095705004 @default.
- W3002961036 cites W2127141656 @default.
- W3002961036 cites W2144499799 @default.
- W3002961036 cites W2183341477 @default.
- W3002961036 cites W2193413348 @default.
- W3002961036 cites W2194775991 @default.
- W3002961036 cites W2327501763 @default.
- W3002961036 cites W2514741789 @default.
- W3002961036 cites W2739883972 @default.
- W3002961036 cites W2766219058 @default.
- W3002961036 cites W2892009249 @default.
- W3002961036 cites W2913718171 @default.
- W3002961036 cites W2962780374 @default.
- W3002961036 cites W2962824709 @default.
- W3002961036 cites W2963403868 @default.
- W3002961036 cites W2964110616 @default.
- W3002961036 cites W2964272710 @default.
- W3002961036 cites W2964308564 @default.
- W3002961036 cites W2972389417 @default.
- W3002961036 cites W2972451902 @default.
- W3002961036 cites W2972818416 @default.
- W3002961036 cites W2972977747 @default.
- W3002961036 cites W854541894 @default.
- W3002961036 doi "https://doi.org/10.48550/arxiv.2001.08290" @default.
- W3002961036 hasPublicationYear "2020" @default.
- W3002961036 type Work @default.
- W3002961036 sameAs 3002961036 @default.
- W3002961036 citedByCount "0" @default.
- W3002961036 crossrefType "posted-content" @default.
- W3002961036 hasAuthorship W3002961036A5006453801 @default.
- W3002961036 hasAuthorship W3002961036A5027112640 @default.
- W3002961036 hasAuthorship W3002961036A5036446253 @default.
- W3002961036 hasAuthorship W3002961036A5066595006 @default.
- W3002961036 hasAuthorship W3002961036A5072887461 @default.
- W3002961036 hasBestOaLocation W30029610361 @default.
- W3002961036 hasConcept C111919701 @default.
- W3002961036 hasConcept C118505674 @default.
- W3002961036 hasConcept C119599485 @default.
- W3002961036 hasConcept C123657996 @default.
- W3002961036 hasConcept C127413603 @default.
- W3002961036 hasConcept C133488467 @default.
- W3002961036 hasConcept C137293760 @default.
- W3002961036 hasConcept C142362112 @default.
- W3002961036 hasConcept C147168706 @default.
- W3002961036 hasConcept C153349607 @default.
- W3002961036 hasConcept C154945302 @default.
- W3002961036 hasConcept C165801399 @default.
- W3002961036 hasConcept C206588197 @default.
- W3002961036 hasConcept C28490314 @default.
- W3002961036 hasConcept C41008148 @default.
- W3002961036 hasConcept C50644808 @default.
- W3002961036 hasConcept C548081761 @default.
- W3002961036 hasConcept C66322947 @default.
- W3002961036 hasConcept C74296488 @default.
- W3002961036 hasConcept C76155785 @default.
- W3002961036 hasConcept C82876162 @default.
- W3002961036 hasConceptScore W3002961036C111919701 @default.
- W3002961036 hasConceptScore W3002961036C118505674 @default.
- W3002961036 hasConceptScore W3002961036C119599485 @default.
- W3002961036 hasConceptScore W3002961036C123657996 @default.
- W3002961036 hasConceptScore W3002961036C127413603 @default.
- W3002961036 hasConceptScore W3002961036C133488467 @default.
- W3002961036 hasConceptScore W3002961036C137293760 @default.
- W3002961036 hasConceptScore W3002961036C142362112 @default.
- W3002961036 hasConceptScore W3002961036C147168706 @default.
- W3002961036 hasConceptScore W3002961036C153349607 @default.
- W3002961036 hasConceptScore W3002961036C154945302 @default.
- W3002961036 hasConceptScore W3002961036C165801399 @default.
- W3002961036 hasConceptScore W3002961036C206588197 @default.
- W3002961036 hasConceptScore W3002961036C28490314 @default.
- W3002961036 hasConceptScore W3002961036C41008148 @default.
- W3002961036 hasConceptScore W3002961036C50644808 @default.
- W3002961036 hasConceptScore W3002961036C548081761 @default.
- W3002961036 hasConceptScore W3002961036C66322947 @default.
- W3002961036 hasConceptScore W3002961036C74296488 @default.
- W3002961036 hasConceptScore W3002961036C76155785 @default.
- W3002961036 hasConceptScore W3002961036C82876162 @default.
- W3002961036 hasLocation W30029610361 @default.
- W3002961036 hasOpenAccess W3002961036 @default.
- W3002961036 hasPrimaryLocation W30029610361 @default.
- W3002961036 hasRelatedWork W11798771 @default.
- W3002961036 hasRelatedWork W11920722 @default.
- W3002961036 hasRelatedWork W12168553 @default.
- W3002961036 hasRelatedWork W14230040 @default.
- W3002961036 hasRelatedWork W14754660 @default.
- W3002961036 hasRelatedWork W1745277 @default.
- W3002961036 hasRelatedWork W2308727 @default.
- W3002961036 hasRelatedWork W3939953 @default.
- W3002961036 hasRelatedWork W1226999 @default.