Data-based melody generation through multi-objective evolutionary computation

Abstract

Genetic-based composition algorithms are able to explore an immense space of possibilities, but the main difficulty has always been the implementation of the selection process. In this work, sets of melodies are utilized for training a machine learning approach to compute fitness, based on different metrics. The fitness of a candidate is provided by combining the metrics, but their values can range through different orders of magnitude and evolve in different ways, which makes it hard to combine these criteria. In order to solve this problem, a multi-objective fitness approach is proposed, in which the best individuals are those in the Pareto-optimal frontier of the multi-dimensional fitness space. Melodic trees are also proposed as a data structure for chromosomic representation of melodies and genetic operators are adapted to them. Some experiments have been carried out using a graphical interface prototype that allows one to explore the creative capabilities of the proposed system. An Online Supplement is provided where the reader can find some technical details, information about the data used, generated melodies, and additional information about the developed prototype and its performance.

Applications of Computational Intelligence in Computer Music Composition

“…Their objective was to obtain the probabilities of a given note following the last note incorporated in the melody. Pedro et al [7] proposed a multi-objective fitness function genetic algorithm (GA) for enhancing the selection process of melody generation. Moreover, they proposed "Melodic Trees" as a data structure for chromosomes representation.…”

“…Techniques and Applications Melody Pitch GA: 2016 [7] -CBR: 2017 [6] -DNN: 2016 [8], 2016 [9], 2017 [10] -Markov: 2017 [6] -GAN: 2017 [11], 2018 [12], 2020 [14], 2021 […”

Evolving Musical Sight Reading Exercises Using Expert Models

“…The trees proposed by Rizo et al (2003) andde León et al (2016) both support simple time signatures-that is, time signatures such as 4 4 and 2 4 where measures can recursively be divided into equal halves without the need for dotted notes. However, neither of these trees support compound time signatures such as 6 8 and 9 8 , where measures do not neatly fit into a binary structure.…”

“…A time signature is applied to the melody when translating it to a score rather than within the tree itself, and the melody has no guarantee of fitting within the chosen time signature. This means that Dahlstedt's tree, unlike the trees of Rizo et al (2003) andde León et al (2016), also does not maintain a musical grammar. That is, it can not FIGURE 2 | The duration hierarchy typically used by tree structures which are designed to represent melodies in common time.…”

“…Triplets are implemented with an internal node operator similar to the split and continuation operators used by de León et al (2016). As shown in Figure 3, the triplet operator is placed on the first direct parent of the triplet leaf nodes.…”