A customized residual neural network and bi-directional gated recurrent unit-based automatic speech recognition model
Autor(a) principal: | |
---|---|
Data de Publicação: | 2022 |
Outros Autores: | , , |
Tipo de documento: | Artigo |
Idioma: | eng |
Título da fonte: | Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) |
Texto Completo: | https://hdl.handle.net/10216/145993 |
Resumo: | Speech recognition aims to convert human speech into text and has applications in security, healthcare, commerce, automobiles, and technology, just to name a few. Inserting residual neural networks before recurrent neural network cells improves accuracy and cuts training time by a good margin. Furthermore, layer normalization instead of batch normalization is more effective in model training and performance enhancement. Also, the size of the datasets presents tremendous influences in achieving the best performance. Leveraging these tricks, this article proposes an automatic speech recognition model with a stacked five layers of customized Residual Convolution Neural Network and seven layers of Bi-Directional Gated Recurrent Units, including a logarithmic so f tmax for the model output. Each of them incorporates a learnable per-element affine parameter-based layer normalization technique. The training and testing of the new model were conducted on the LibriSpeech corpus and LJ Speech dataset. The experimental results demonstrate a character error rate (CER) of 4.7 and 3.61% on the two datasets, respectively, with only 33 million parameters without the requirement of any external language model. |
id |
RCAP_c75525e1280371c258e61e641729d9c5 |
---|---|
oai_identifier_str |
oai:repositorio-aberto.up.pt:10216/145993 |
network_acronym_str |
RCAP |
network_name_str |
Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) |
repository_id_str |
7160 |
spelling |
A customized residual neural network and bi-directional gated recurrent unit-based automatic speech recognition modelCiências Tecnológicas, Ciências da engenharia e tecnologiasTechnological sciences, Engineering and technologySpeech recognition aims to convert human speech into text and has applications in security, healthcare, commerce, automobiles, and technology, just to name a few. Inserting residual neural networks before recurrent neural network cells improves accuracy and cuts training time by a good margin. Furthermore, layer normalization instead of batch normalization is more effective in model training and performance enhancement. Also, the size of the datasets presents tremendous influences in achieving the best performance. Leveraging these tricks, this article proposes an automatic speech recognition model with a stacked five layers of customized Residual Convolution Neural Network and seven layers of Bi-Directional Gated Recurrent Units, including a logarithmic so f tmax for the model output. Each of them incorporates a learnable per-element affine parameter-based layer normalization technique. The training and testing of the new model were conducted on the LibriSpeech corpus and LJ Speech dataset. The experimental results demonstrate a character error rate (CER) of 4.7 and 3.61% on the two datasets, respectively, with only 33 million parameters without the requirement of any external language model.2022-042022-04-01T00:00:00Zinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/articleapplication/pdfimage/pnghttps://hdl.handle.net/10216/145993eng0957-417410.1016/j.eswa.2022.119293Selim RezaMarta Campos FerreiraJ.J.M. MachadoJoão Manuel R. S. Tavaresinfo:eu-repo/semantics/openAccessreponame:Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)instname:Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informaçãoinstacron:RCAAP2023-11-29T13:46:47Zoai:repositorio-aberto.up.pt:10216/145993Portal AgregadorONGhttps://www.rcaap.pt/oai/openaireopendoar:71602024-03-19T23:47:24.327412Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) - Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informaçãofalse |
dc.title.none.fl_str_mv |
A customized residual neural network and bi-directional gated recurrent unit-based automatic speech recognition model |
title |
A customized residual neural network and bi-directional gated recurrent unit-based automatic speech recognition model |
spellingShingle |
A customized residual neural network and bi-directional gated recurrent unit-based automatic speech recognition model Selim Reza Ciências Tecnológicas, Ciências da engenharia e tecnologias Technological sciences, Engineering and technology |
title_short |
A customized residual neural network and bi-directional gated recurrent unit-based automatic speech recognition model |
title_full |
A customized residual neural network and bi-directional gated recurrent unit-based automatic speech recognition model |
title_fullStr |
A customized residual neural network and bi-directional gated recurrent unit-based automatic speech recognition model |
title_full_unstemmed |
A customized residual neural network and bi-directional gated recurrent unit-based automatic speech recognition model |
title_sort |
A customized residual neural network and bi-directional gated recurrent unit-based automatic speech recognition model |
author |
Selim Reza |
author_facet |
Selim Reza Marta Campos Ferreira J.J.M. Machado João Manuel R. S. Tavares |
author_role |
author |
author2 |
Marta Campos Ferreira J.J.M. Machado João Manuel R. S. Tavares |
author2_role |
author author author |
dc.contributor.author.fl_str_mv |
Selim Reza Marta Campos Ferreira J.J.M. Machado João Manuel R. S. Tavares |
dc.subject.por.fl_str_mv |
Ciências Tecnológicas, Ciências da engenharia e tecnologias Technological sciences, Engineering and technology |
topic |
Ciências Tecnológicas, Ciências da engenharia e tecnologias Technological sciences, Engineering and technology |
description |
Speech recognition aims to convert human speech into text and has applications in security, healthcare, commerce, automobiles, and technology, just to name a few. Inserting residual neural networks before recurrent neural network cells improves accuracy and cuts training time by a good margin. Furthermore, layer normalization instead of batch normalization is more effective in model training and performance enhancement. Also, the size of the datasets presents tremendous influences in achieving the best performance. Leveraging these tricks, this article proposes an automatic speech recognition model with a stacked five layers of customized Residual Convolution Neural Network and seven layers of Bi-Directional Gated Recurrent Units, including a logarithmic so f tmax for the model output. Each of them incorporates a learnable per-element affine parameter-based layer normalization technique. The training and testing of the new model were conducted on the LibriSpeech corpus and LJ Speech dataset. The experimental results demonstrate a character error rate (CER) of 4.7 and 3.61% on the two datasets, respectively, with only 33 million parameters without the requirement of any external language model. |
publishDate |
2022 |
dc.date.none.fl_str_mv |
2022-04 2022-04-01T00:00:00Z |
dc.type.status.fl_str_mv |
info:eu-repo/semantics/publishedVersion |
dc.type.driver.fl_str_mv |
info:eu-repo/semantics/article |
format |
article |
status_str |
publishedVersion |
dc.identifier.uri.fl_str_mv |
https://hdl.handle.net/10216/145993 |
url |
https://hdl.handle.net/10216/145993 |
dc.language.iso.fl_str_mv |
eng |
language |
eng |
dc.relation.none.fl_str_mv |
0957-4174 10.1016/j.eswa.2022.119293 |
dc.rights.driver.fl_str_mv |
info:eu-repo/semantics/openAccess |
eu_rights_str_mv |
openAccess |
dc.format.none.fl_str_mv |
application/pdf image/png |
dc.source.none.fl_str_mv |
reponame:Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) instname:Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação instacron:RCAAP |
instname_str |
Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação |
instacron_str |
RCAAP |
institution |
RCAAP |
reponame_str |
Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) |
collection |
Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) |
repository.name.fl_str_mv |
Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) - Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação |
repository.mail.fl_str_mv |
|
_version_ |
1799135791921758208 |