Carlos De La Vega Martin

Profile

Project title:

Neural Drum Synthesis

Abstract:

Drum sound synthesis is a prominent part of modern music production. Two different approaches are commonly used nowadays, abstract (no underlying physical interpretation) techniques such as subtractive or additive synthesis, and sample-based synthesizers.

Physical modelling
Physical modelling, in particular employing finite difference schemes, has been explored to overcome the limitations of the previously discussed synthesis techniques. Physical modelling gives a real-world interpretation to synthesis parameters, as well as extensive control over them.

Neural Networks
In the last 20 years, neural networks have become a widely used method for solving complex, highly non-linear problems. Due to their properties as universal function approximators, there has been a growing interest in using neural networks for solving systems of partial differential equations.

C4DM theme affiliation:

Sound Synthesis

Research