Neural network assisted composition for piano in jazz

Ismael Medina Muñoz (2023)

Artificial Intelligence has taken an important role in activities that were once considered exclusively human. Generative AI is a vibrant area of research, with increasing interest in application fields related to the arts. The recent plethora of innovations in fields like visual arts and natural language processing, which are able to engage in dialogue with users, are just two examples of commercial applications that are driving innovation research for big tech giants. It would not be untrue to say that these innovations are shaping mankind’s development.

Music is an investigative field that presents a challenge. Musical theory itself is challenging for humans, and music is as diverse and rich as the cultures in which it has evolved. This research and proposal is intended as a novel approach to creating a generative artificial intelligence that assists in piano composition for jazz tunes. This genre was selected because of the challenge that its richness and complexity for musical execution and interpretation pose.

By using a Recurrent Neural Net to create new sequences of n-notes from an initial n-note set and using a probabilistic approach to set the duration of each note in the produced n-notes set, the generative artificial intelligence described in this document is the piano composer assistant for jazz tunes.

Tesis de maestría

Inteligencia Artificial INGENIERÍA Y TECNOLOGÍA CIENCIAS TECNOLÓGICAS TECNOLOGÍA DE LAS TELECOMUNICACIONES DISPOSITIVOS DE GRABACIÓN DISPOSITIVOS DE GRABACIÓN