Virtual analog modelling using machine learning

Abstract

Digital reproduction of the behaviour of analog electrical circuits such as synth filters or guitar amplifiers has been a common topic in music signal processing research since the mid-1990s. In the last several years the modern techniques of machine learning have started to be applied to this problem, producing very promising results. In parallel, the same time period has seen an explosion of work in the application of machine learning to solution of differential equations. This talk will give an overview of work in these closely connected fields, and explore how different application areas have influenced the type of solutions developed.

Bio

Julian Parker
Native Instruments
Google Scholar | Github

Julian Parker is the Principal Software Engineer for Machine Learning and DSP at Native Instruments GmbH. He received his bachelors degree in Natural Sciences from the University of Cambridge in 2005, and an M.Sc. in Acoustics & Music Technology from the University of Edinburgh in 2008. In 2013, he completed his doctoral degree at Aalto University, Finland. Since 2013 he has been employed at Native Instruments, where he has worked on a wide range of products in the sound synthesis and effects domain. He now leads DSP and audio-focused ML research and development at the company. Julian is an active member of the academic community and has published on a variety of topics including reverberation, physical modelling, digital filter design and machine learning.