Harmonização musical automática baseada em redes neurais artificiais
In musical terms, harmony is defined as the art of combining several musical notes simultaneously to create a coherent sound to human ears, with the purpose of accompanying, framing and filling. However, the elaboration of a harmony for a particular melody is not a trivial task, since it requires mu...
| Autor principal: | Costa, Lucas Francesco Piccioni |
|---|---|
| Formato: | Trabalho de Conclusão de Curso (Graduação) |
| Idioma: | Português |
| Publicado em: |
Universidade Tecnológica Federal do Paraná
2020
|
| Assuntos: | |
| Acesso em linha: |
http://repositorio.utfpr.edu.br/jspui/handle/1/15792 |
| Tags: |
Adicionar Tag
Sem tags, seja o primeiro a adicionar uma tag!
|
| Resumo: |
In musical terms, harmony is defined as the art of combining several musical notes simultaneously to create a coherent sound to human ears, with the purpose of accompanying, framing and filling. However, the elaboration of a harmony for a particular melody is not a trivial task, since it requires music knowledge and experience, as well as intense study of music theory, which takes time and may even take years to reach reasonable skills. Thus, the development of a system capable of automatically harmonizing melodies is beneficial. Efforts and attempts have been made in solving the harmonization problem, as shown in the method proposed by Koops, Magalhães e Haas (2013), which generates chords for each note of a given melody and selects the best sequence by matching it to the input melody in one. single output file. In the present project, it was proposed to build a tool for the harmonization of predefined melodies, using artificial intelligence techniques with results that are realistic, consistent with harmony theory rules and in order to present diversity. Thus, a melody defined with symbolic and properly processed musical data serves as input to an artificial neural network generating for this chords, having been trained based on a music database containing melody and harmony. The system output is the chord sequence to be applied to the melody. Due to the complexity of obtaining and analyzing results, quantitative and qualitative tests were performed, using from performance measures calculated from confusion matrices to online polling with anonymous volunteer participants opinion. In the end, it can be concluded that the system was able to generate harmonies aligned with the objectives. |
|---|