معرفی
David Südholt is a Visiting Professor at the School of Electronic Engineering and Computer Science, Queen Mary University of London, affiliated with the Centre for Digital Music (C4DM). His research focuses on machine learning integration with physical models for voice synthesis, exploring applications in text-to-speech systems, singing voice synthesis, and unconventional vocal expressions. He collaborates with industry partner Nemisindo and investigates themes like audio engineering and sound synthesis. His work bridges machine learning techniques with traditional physical modeling approaches to enhance synthesis quality and expressive capabilities.
Key research interests include vocal tract modeling, timbre manipulation, real-time signal processing, and the implementation of finite difference schemes in audio synthesis. His projects aim to expand the expressive range of physical models to include non-standard vocal techniques (e.g., screams, whispers) through machine learning-driven parameter estimation.
Recent publications highlight advancements in gradient-based vocal tract estimation, differentiable digital signal processing, and real-time timbre transfer using frameworks like DDSP. He also contributes to FAUST-based implementations of audio algorithms and interactive web-based sound synthesis tools for instruments like the langeleik.
No academic awards or student advisement records are listed in the provided information.
