Survey: Style Replication

Abstract

The aim of the present work is to perform a step towards the design of specific algorithms and methods for automatic music generation. A novel probabilistic model for the characterization of music learned from music samples is designed. This model makes use of automatically extracted music parameters, namely tempo, time signature, tonality, rhythmic patterns, harmonic evolution and pitch contour to characterize musical styles. Then, a novel autonomous music compositor that generates new melodies using the model developed will be presented. The methods proposed in this paper take into consideration different aspects related to the traditional way in which music is composed by humans such as harmony evolution and structure repetitions and apply them together with the probabilistic reutilization of rhythm patterns and pitch contours learned beforehand to compose music pieces.

What is this survey about?

Next, several groups of audio samples are going to presented. In each group, one music sample will represent the training data used for the style and composition modeling, while the other three will represent the generated score using the proposed system.

The aim of this survey, is to evaluate if the style replication process is successfully performed.

Each participant is asked to score from 1 to 5 (where 1 means not similar and 5 very similar) the similarity of style between the training set and the generated set.

Group 1: Pop

Training Set Generated Set

(Not Similar) 1 2 3 4 5 (Very Similar)

Group 2: Academic

Training Set Generated Set

(Not Similar) 1 2 3 4 5 (Very Similar)

Group 3: Flamenco

Training Set Generated Set

(Not Similar) 1 2 3 4 5 (Very Similar)

If you want to leave any comment or suggestion, please feel free to write it below:

If you have any issue with the survey, please contact to carles@ic.uma.es