Título: Análise de Desempenho de Modelos de Regressão Entre Avaliações Subjetivas e Objetivas Aplicados a Sistemas de Baixo Virtual
Autores: Danilo O. Carvalho, Carmelo J. A. Bastos-Filho, Sergio C. Oliveira
Resumo: Due to their physical constraints, small-size loudspeakers cannot reproduce low frequencies free of unexpected effects such as distortion. In this article, a psycho-acoustic phenomena-based system was developed to process bass sounds in tracks with different features. The resulting signals are assessed by a set of volunteers and an objective algorithm. Based on these data, some regression processes (linear, gaussian, neural network, regression trees for exemple) are used to provide functions to generate similar volunteer scores. Validation experiments were performed considering the perception of human beings and automatic methods using 5 songs with different characteristics to evaluate the regression process using a speaker with low frequency reproduction restrictions. Their performance numbers are compared and presented to validate the effectiveness of the system, where gaussian regretion models lead to better results
Palavras-chave: nan
Páginas: 8
Código DOI: 10.21528/CBIC2023-087
Artigo em pdf: CBIC_2023_paper087.pdf
Arquivo BibTeX: CBIC_2023_087.bib