| Popis: |
Drum pattern generation is a task that focuses on the rhythmic aspect of music and aims at generating percussive sequences. With the advancement of machine learning techniques, several models have been proven useful in producing compelling results. However, one of the main challenges is to generate structurally cohesive sequences. In this study, a drum pattern generation model based on Variational Autoencoders (VAEs) is presented; Specifically, the proposed model is built to generate symbolic drum patterns given an accompaniment that consists of melodic sequences. A self-similarity matrix (SSM) is incorporated in the process for encapsulating structural information. Both the objective evaluation and the subjective listening test highlight the model's capability of creating musically meaningful transitions on structural boundaries. |
