A probabilistically-oriented analysis of the performance of ASR Systems for brazilian radios and TVs

Azevedo, Diego Marques de; Rodrigues, Guilherme Souza; Ladeira, Marcelo

Registro completo de metadados

Campo DC	Valor	Idioma
dc.contributor.author	Azevedo, Diego Marques de	-
dc.contributor.author	Rodrigues, Guilherme Souza	-
dc.contributor.author	Ladeira, Marcelo	-
dc.date.accessioned	2023-10-11T20:07:51Z	-
dc.date.available	2023-10-11T20:07:51Z	-
dc.date.issued	2022-11-19	-
dc.identifier.citation	AZEVEDO, Diego Marques de; RODRIGUES, Guilherme Souza; LADEIRA, Marcelo. A probabilistically-oriented analysis of the performance of ASR Systems for brazilian radios and TVs. In: BRACIS, 11., 2022, Campinas. Anais [...]. Campinas: Springer, 2022. p. 169-180. DOI: https://doi.org/10.1007/978-3-031-21689-3_13. Disponível em: https://link.springer.com/chapter/10.1007/978-3-031-21689-3_13. Acesso em: 11 out. 2023.	pt_BR
dc.identifier.uri	http://repositorio2.unb.br/jspui/handle/10482/46671	-
dc.language.iso	eng	pt_BR
dc.publisher	Springer	pt_BR
dc.rights	Acesso Restrito	pt_BR
dc.title	A probabilistically-oriented analysis of the performance of ASR Systems for brazilian radios and TVs	pt_BR
dc.type	Trabalho apresentado em evento	pt_BR
dc.subject.keyword	Reconhecimento automático da voz	pt_BR
dc.subject.keyword	Redes neurais (Computação)	pt_BR
dc.subject.keyword	Inteligência artificial	pt_BR
dc.subject.keyword	Áudio - propriedades	pt_BR
dc.rights.license	© 2022 The Author(s), under exclusive license to Springer Nature Switzerland AG.	pt_BR
dc.identifier.doi	https://doi.org/10.1007/978-3-031-21689-3_13	pt_BR
dc.description.abstract1	With the use of neural network-based technologies, Automatic Speech Recognition (ASR) systems for Brazilian Portuguese (BP) have shown great progress in the last few years. Several state-of-art results were achieved by open-source end-to-end models, such as the Kaldi toolkit and the Wav2vec 2.0. Alternative commercial tools are also available, including the Google and Microsoft speech to text APIs and the Audimus System of VoiceInteraction. We analyse the relative performance of such tools – in terms of the so-called Word Error Rate (WER) – when transcribing audio recordings from Brazilian radio and TV channels. A generalized linear model (GLM) is designed to stochastically describe the relationship between some of the audio’s properties (e.g. file format and audio duration) and the resulting WER, for each method under consideration. Among other uses, such strategy enables the analysis of local performances, indicating not only which tool performs better, but when exactly it is expected to do so. This, in turn, could be used to design an optimized system composed of several transcribers. The data generated for conducting this experiment and the scripts used to produce the stochastic model are public available.	pt_BR
dc.contributor.affiliation	University of Brasília, Instituto de Ciências Exatas, Departamento de Estatística	pt_BR
dc.contributor.affiliation	University of Brasília, Instituto de Ciências Exatas, Departamento de Estatística	pt_BR
dc.contributor.affiliation	University of Brasília, Instituto de Ciências Exatas, Departamento de Ciência da Computação	pt_BR
dc.description.unidade	Instituto de Ciências Exatas (IE)	pt_BR
dc.description.unidade	Departamento de Ciência da Computação (IE CIC)	pt_BR
dc.description.unidade	Departamento de Estatística (IE EST)	pt_BR
dc.description.ppg	Programa de Pós-Graduação em Computação Aplicada, Mestrado Profissional	pt_BR
Aparece nas coleções:	Trabalhos apresentados em evento