Redes neuronales recurrentes para la generación automática de música

Cea Morán, Juan Julián (2020). Redes neuronales recurrentes para la generación automática de música. Thesis (Master thesis), E.T.S. de Ingenieros Informáticos (UPM).

Description

Title: Redes neuronales recurrentes para la generación automática de música
Author/s:
  • Cea Morán, Juan Julián
Contributor/s:
  • Serradilla García, Francisco J.
Item Type: Thesis (Master thesis)
Masters title: Inteligencia Artificial
Date: 2020
Subjects:
Faculty: E.T.S. de Ingenieros Informáticos (UPM)
Department: Inteligencia Artificial
Creative Commons Licenses: Recognition - No derivative works - Non commercial

Full text

[img]
Preview
PDF - Requires a PDF viewer, such as GSview, Xpdf or Adobe Acrobat Reader
Download (790kB) | Preview

Abstract

Usualmente se tiende a pensar en los sistemas informáticos como elementos puramente objetivos y deterministas, capaces de generar cierta salida dada cierta entrada. Dentro del campo de la Inteligencia Artificial, se pretenden resolver problemas del mismo modo que lo haría un humano, y en muchas ocasiones eso requiere que las técnicas y herramientas usadas sean puramente objetivas. Sin embargo, existen problemáticas que no pueden ser resueltas de esta forma, si no que tal y como lo harían los humanos, necesitan de cierta creatividad. El problema de la generación musical automática, es uno de estos supuestos en los que se desea que el sistema que lo resuelva, sea capaz de plasmar cierta creatividad. Afortunadamente existen una gran variedad de algoritmos y métodos que permiten realizar esta tarea. Dentro del campo del Deep Learning, existe un tipo de arquitectura neuronal que destaca por encima del resto: las Redes Neuronales Recurrentes. En esta tesis se pretende entender este tipo de arquitecturas, explicando por qué son tan populares en el campo de la composición musical automática. Además, se implementan algunas topologías con el fin de ejemplificar lo aprendido y demostrar que efectivamente existen modelos capaces de aportar soluciones creativas a problemas humanos.---ABSTRACT---There is usually a tendency to think of computer systems as purely objective and deterministic elements, capable of generating a certain output given a certain input. Within the field of Artificial Intelligence, the aim is to solve problems in the same way that a human would, and on many occasions this requires the techniques and tools used to be purely objective. However, there are problems that cannot be solved in this way, but rather just as humans would, they need a certain creativity. The problem of automatic music generation is an example in which we want the system that solves it to be able to show some creativity. Fortunately, there is a great variety of algorithms and methods that allow us to carry out this task. In the Deep Learning eld, there is a type of neuronal architecture that stands out from the rest: the Recurrent Neuronal Networks. This thesis aims to understand this type of architecture, explaining why they are so popular for automatic music composition. In addition, some topologies are implemented in order to exemplify what has been learned and demonstrate that there are indeed algorithms capable of providing creative solutions to human problems.

More information

Item ID: 63687
DC Identifier: http://oa.upm.es/63687/
OAI Identifier: oai:oa.upm.es:63687
Deposited by: Biblioteca Facultad de Informatica
Deposited on: 09 Sep 2020 11:13
Last Modified: 09 Sep 2020 11:13
  • Logo InvestigaM (UPM)
  • Logo GEOUP4
  • Logo Open Access
  • Open Access
  • Logo Sherpa/Romeo
    Check whether the anglo-saxon journal in which you have published an article allows you to also publish it under open access.
  • Logo Dulcinea
    Check whether the spanish journal in which you have published an article allows you to also publish it under open access.
  • Logo de Recolecta
  • Logo del Observatorio I+D+i UPM
  • Logo de OpenCourseWare UPM