CORAA ASR: a large corpus of spontaneous and prepared speech manually validated for speech recognition in Brazilian Portuguese
Autor(a) principal: | |
---|---|
Data de Publicação: | 2022 |
Outros Autores: | , , , , , , , , , |
Tipo de documento: | Artigo |
Idioma: | eng |
Título da fonte: | Repositório Institucional da UNESP |
Texto Completo: | http://dx.doi.org/10.1007/s10579-022-09621-4 http://hdl.handle.net/11449/246341 |
Resumo: | Automatic Speech recognition (ASR) is a complex and challenging task. In recent years, there have been significant advances in the area. In particular, for the Brazilian Portuguese (BP) language, there were around 376 h publicly available for the ASR task until the second half of 2020. With the release of new datasets in early 2021, this number increased to 574 h. The existing resources, however, are composed of audios containing only read and prepared speech. There is a lack of datasets including spontaneous speech, which are essential in several ASR applications. This paper presents CORAA (Corpus of Annotated Audios) ASR with 290 h, a publicly available dataset for ASR in BP containing validated pairs of audio-transcription. CORAA ASR also contains European Portuguese audios (4.6 h). We also present a public ASR model based on Wav2Vec 2.0 XLSR-53, fine-tuned over CORAA ASR. Our model achieved a Word Error Rate (WER) of 24.18% on CORAA ASR test set and 20.08% on Common Voice test set. When measuring the Character Error Rate (CER), we obtained 11.02% and 6.34% for CORAA ASR and Common Voice, respectively. CORAA ASR corpora were assembled to both improve ASR models in BP with phenomena from spontaneous speech and motivate young researchers to start their studies on ASR for Portuguese. All the corpora are publicly available at https://github.com/nilc-nlp/CORAA under the CC BY-NC-ND 4.0 license. |
id |
UNSP_d1dd81e8951e741342b0135274dd63bc |
---|---|
oai_identifier_str |
oai:repositorio.unesp.br:11449/246341 |
network_acronym_str |
UNSP |
network_name_str |
Repositório Institucional da UNESP |
repository_id_str |
2946 |
spelling |
CORAA ASR: a large corpus of spontaneous and prepared speech manually validated for speech recognition in Brazilian PortugueseAutomatic speech recognitionBrazilian PortuguesePrepared speechPublic datasetsPublic speech corporaSpontaneous speechAutomatic Speech recognition (ASR) is a complex and challenging task. In recent years, there have been significant advances in the area. In particular, for the Brazilian Portuguese (BP) language, there were around 376 h publicly available for the ASR task until the second half of 2020. With the release of new datasets in early 2021, this number increased to 574 h. The existing resources, however, are composed of audios containing only read and prepared speech. There is a lack of datasets including spontaneous speech, which are essential in several ASR applications. This paper presents CORAA (Corpus of Annotated Audios) ASR with 290 h, a publicly available dataset for ASR in BP containing validated pairs of audio-transcription. CORAA ASR also contains European Portuguese audios (4.6 h). We also present a public ASR model based on Wav2Vec 2.0 XLSR-53, fine-tuned over CORAA ASR. Our model achieved a Word Error Rate (WER) of 24.18% on CORAA ASR test set and 20.08% on Common Voice test set. When measuring the Character Error Rate (CER), we obtained 11.02% and 6.34% for CORAA ASR and Common Voice, respectively. CORAA ASR corpora were assembled to both improve ASR models in BP with phenomena from spontaneous speech and motivate young researchers to start their studies on ASR for Portuguese. All the corpora are publicly available at https://github.com/nilc-nlp/CORAA under the CC BY-NC-ND 4.0 license.Federal University of Technology — Paraná (UTFPR)Instituto de Ciências Matemáticas e de Computação - University of São PauloFederal University of GoiasSão Paulo State UniversitySão Paulo State UniversityFederal University of Technology — Paraná (UTFPR)Universidade de São Paulo (USP)Federal University of GoiasUniversidade Estadual Paulista (UNESP)Candido Junior, Arnaldo [UNESP]Casanova, EdressonSoares, Andersonde Oliveira, Frederico SantosOliveira, LucasJunior, Ricardo Corso Fernandesda Silva, Daniel Peixoto PintoFayet, Fernando GorgulhoCarlotto, Bruno BaldisseraGris, Lucas Rafael StefanelAluísio, Sandra Maria2023-07-29T12:38:19Z2023-07-29T12:38:19Z2022-01-01info:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/articlehttp://dx.doi.org/10.1007/s10579-022-09621-4Language Resources and Evaluation.1572-84121574-020Xhttp://hdl.handle.net/11449/24634110.1007/s10579-022-09621-42-s2.0-85142235932Scopusreponame:Repositório Institucional da UNESPinstname:Universidade Estadual Paulista (UNESP)instacron:UNESPengLanguage Resources and Evaluationinfo:eu-repo/semantics/openAccess2023-07-29T12:38:19Zoai:repositorio.unesp.br:11449/246341Repositório InstitucionalPUBhttp://repositorio.unesp.br/oai/requestopendoar:29462024-08-05T14:26:05.451253Repositório Institucional da UNESP - Universidade Estadual Paulista (UNESP)false |
dc.title.none.fl_str_mv |
CORAA ASR: a large corpus of spontaneous and prepared speech manually validated for speech recognition in Brazilian Portuguese |
title |
CORAA ASR: a large corpus of spontaneous and prepared speech manually validated for speech recognition in Brazilian Portuguese |
spellingShingle |
CORAA ASR: a large corpus of spontaneous and prepared speech manually validated for speech recognition in Brazilian Portuguese Candido Junior, Arnaldo [UNESP] Automatic speech recognition Brazilian Portuguese Prepared speech Public datasets Public speech corpora Spontaneous speech |
title_short |
CORAA ASR: a large corpus of spontaneous and prepared speech manually validated for speech recognition in Brazilian Portuguese |
title_full |
CORAA ASR: a large corpus of spontaneous and prepared speech manually validated for speech recognition in Brazilian Portuguese |
title_fullStr |
CORAA ASR: a large corpus of spontaneous and prepared speech manually validated for speech recognition in Brazilian Portuguese |
title_full_unstemmed |
CORAA ASR: a large corpus of spontaneous and prepared speech manually validated for speech recognition in Brazilian Portuguese |
title_sort |
CORAA ASR: a large corpus of spontaneous and prepared speech manually validated for speech recognition in Brazilian Portuguese |
author |
Candido Junior, Arnaldo [UNESP] |
author_facet |
Candido Junior, Arnaldo [UNESP] Casanova, Edresson Soares, Anderson de Oliveira, Frederico Santos Oliveira, Lucas Junior, Ricardo Corso Fernandes da Silva, Daniel Peixoto Pinto Fayet, Fernando Gorgulho Carlotto, Bruno Baldissera Gris, Lucas Rafael Stefanel Aluísio, Sandra Maria |
author_role |
author |
author2 |
Casanova, Edresson Soares, Anderson de Oliveira, Frederico Santos Oliveira, Lucas Junior, Ricardo Corso Fernandes da Silva, Daniel Peixoto Pinto Fayet, Fernando Gorgulho Carlotto, Bruno Baldissera Gris, Lucas Rafael Stefanel Aluísio, Sandra Maria |
author2_role |
author author author author author author author author author author |
dc.contributor.none.fl_str_mv |
Federal University of Technology — Paraná (UTFPR) Universidade de São Paulo (USP) Federal University of Goias Universidade Estadual Paulista (UNESP) |
dc.contributor.author.fl_str_mv |
Candido Junior, Arnaldo [UNESP] Casanova, Edresson Soares, Anderson de Oliveira, Frederico Santos Oliveira, Lucas Junior, Ricardo Corso Fernandes da Silva, Daniel Peixoto Pinto Fayet, Fernando Gorgulho Carlotto, Bruno Baldissera Gris, Lucas Rafael Stefanel Aluísio, Sandra Maria |
dc.subject.por.fl_str_mv |
Automatic speech recognition Brazilian Portuguese Prepared speech Public datasets Public speech corpora Spontaneous speech |
topic |
Automatic speech recognition Brazilian Portuguese Prepared speech Public datasets Public speech corpora Spontaneous speech |
description |
Automatic Speech recognition (ASR) is a complex and challenging task. In recent years, there have been significant advances in the area. In particular, for the Brazilian Portuguese (BP) language, there were around 376 h publicly available for the ASR task until the second half of 2020. With the release of new datasets in early 2021, this number increased to 574 h. The existing resources, however, are composed of audios containing only read and prepared speech. There is a lack of datasets including spontaneous speech, which are essential in several ASR applications. This paper presents CORAA (Corpus of Annotated Audios) ASR with 290 h, a publicly available dataset for ASR in BP containing validated pairs of audio-transcription. CORAA ASR also contains European Portuguese audios (4.6 h). We also present a public ASR model based on Wav2Vec 2.0 XLSR-53, fine-tuned over CORAA ASR. Our model achieved a Word Error Rate (WER) of 24.18% on CORAA ASR test set and 20.08% on Common Voice test set. When measuring the Character Error Rate (CER), we obtained 11.02% and 6.34% for CORAA ASR and Common Voice, respectively. CORAA ASR corpora were assembled to both improve ASR models in BP with phenomena from spontaneous speech and motivate young researchers to start their studies on ASR for Portuguese. All the corpora are publicly available at https://github.com/nilc-nlp/CORAA under the CC BY-NC-ND 4.0 license. |
publishDate |
2022 |
dc.date.none.fl_str_mv |
2022-01-01 2023-07-29T12:38:19Z 2023-07-29T12:38:19Z |
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 |
http://dx.doi.org/10.1007/s10579-022-09621-4 Language Resources and Evaluation. 1572-8412 1574-020X http://hdl.handle.net/11449/246341 10.1007/s10579-022-09621-4 2-s2.0-85142235932 |
url |
http://dx.doi.org/10.1007/s10579-022-09621-4 http://hdl.handle.net/11449/246341 |
identifier_str_mv |
Language Resources and Evaluation. 1572-8412 1574-020X 10.1007/s10579-022-09621-4 2-s2.0-85142235932 |
dc.language.iso.fl_str_mv |
eng |
language |
eng |
dc.relation.none.fl_str_mv |
Language Resources and Evaluation |
dc.rights.driver.fl_str_mv |
info:eu-repo/semantics/openAccess |
eu_rights_str_mv |
openAccess |
dc.source.none.fl_str_mv |
Scopus reponame:Repositório Institucional da UNESP instname:Universidade Estadual Paulista (UNESP) instacron:UNESP |
instname_str |
Universidade Estadual Paulista (UNESP) |
instacron_str |
UNESP |
institution |
UNESP |
reponame_str |
Repositório Institucional da UNESP |
collection |
Repositório Institucional da UNESP |
repository.name.fl_str_mv |
Repositório Institucional da UNESP - Universidade Estadual Paulista (UNESP) |
repository.mail.fl_str_mv |
|
_version_ |
1808128359402569728 |