Perceptually Relevant Models for Articulation in Synthesised Drum Patterns
In this study, we evaluate current techniques for drum pattern humanisation and suggest new methods using a probabilistic model. Our statistical analysis shows that both deviations from a fixed grid and corresponding amplitude values of drum patterns can have non-Gaussian distributions with underlying temporal structures. We plot distributions and probability matrices of sequences played by humans in order to demonstrate this. A new method for humanisation with structural preservation is proposed, using a Markov Chain and an Empirical Cumulative Distribution Function (ECDF) in order to weight pseudorandom variables. Finally we demonstrate the perceptual relevance of these methods using paired listening tests.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is temporarily free for AES members.