CASPIAN JOURNAL
MANAGEMENT AND HIGH TECHNOLOGIES
The application of machine learning methods for the task of generating musical compositions
 |
Read | Nikitin N. A., Rozaliev V. L., Orlova Yu. A., Zaboleeva-Zotova A. V. The application of machine learning methods for the task of generating musical compositions // Caspian journal : management and high technologies. — 2018. — №2. — pp. 84-95. |
Nikitin N. A. - post-graduate student, Volgograd State Technical University, 28 Lenin Ave., Volgograd, 400005, Russian Federation, set.enter@mail.ru
Rozaliev V. L. - РЎand. Sci. (Engineering), Assistant Professor, Volgograd State Technical University, 28 Lenin Ave., Volgograd, 400005, Russian Federation, vladimir.rozaliev@gmail.com
Orlova Yu. A. - Cand. Sci. (Pedagogy), Doct. Sci. (Engineering), Assistant Professor, Volgograd State Technical University, 28 Lenin Ave., Volgograd, 400005, Russian Federation, yulia.orlova@gmail.com
Zaboleeva-Zotova A. V. - Doct. Sci. (Engineering), Professor, Volgograd State Technical University, 28 Lenin Ave., Volgograd, 400005, Russian Federation, zabzot@gmail.com
The purpose of the study described in this article is to increase the harmony and melody of the sound generation based on images. Development of methods that implement a combined approach to the generation of sound sequences. The developed method uses a recurrent neural network to generate musical material and a color music theory, which is used to determine the parameters of a composition by an image. Describes the developed program for generating sounds by the image, based on the developed method, as well as the Python language and the Keras library. The results of experiments that show the high efficiency of the integrated use of methods of machine learning and color music theory for the problem of generating sounds by image are presented.
Key words: рекуррентная нейронная сеть, цветомузыкальная теория, Keras, автоматизированная генерация музыки, схемы соотнесения цветов и нот, анализ изображений, recurrent neural network, color music theory, Keras, automated music generation, color and note matching