C4DM Seminar: Gloria Dal Santo & Sebastian J. Schlecht: Machine Learning-Based Artificial Reverberation & Non-stationary Noise Removal from Repeated sweep Measurements
QMUL, School of Electronic Engineering and Computer Science
Centre for Digital Music Seminar Series
Seminar by:
Gloria Dal Santo & Sebastian J. Schlecht
Date/time: TUesday, 10th September 2024, 1.30pm
**Location: GC601, Graduate Centre Building, Mile End Campus, QMUL, E1 4NS ** Zoom: https://qmul-ac-uk.zoom.us/j/9798452959
Title: Machine Learning-Based Artificial Reverberation & Non-stationary Noise Removal from Repeated sweep Measurements
Abstract: Gloria: The Feedback Delay Network (FDN) is a widely used approach in artificial reverberation, structured by generalizing the parallel comb-filter architecture through the interconnection of delays via a feedback matrix. Motivated by its cost-effectiveness, our research explores the FDN topology and integrates it into a machine learning framework with the aim of defining a novel methodology for accurate real-time room simulation. We are working towards a new paradigm for Room Impulse Response synthesis, where accuracy, cost-effectiveness, and ease of application coexist.
Sebastian: Acoustic measurements using sine sweeps are prone to background noise and non-stationary disturbances. Repeated measurements can be averaged to improve the resulting signal-to-noise ratio. However, averaging leads to poor rejection of non-stationary high-energy disturbances and, in the case of a time-variant environment, causes attenuation at high frequencies. This paper proposes a robust method to combine repeated sweep measurements using across-measurement median filtering in the time-frequency domain. The method, called Mosaic, successfully rejects non-stationary noise, sup- presses background noise and is more robust toward time variation than averaging. The proposed method allows high- quality measurement of impulse responses in a noisy environment.
Bio: Gloria Dal Santo received the B.Sc. degree in Electronic and Communications Engineering from Politecnico di Torino, Turin, Italy, and the M.Sc. degree in Electrical and Electronic Engineering from the Ecole Polytechnique Fédérale de Lausanne, Lausanne, Switzerland, in 2020 and 2022. She is currently working toward the Doctoral degree with the Acoustics Lab, Aalto University, Espoo, Finland. Her research interests include artificial reverberation and audio applications of machine learning.
Sebastian J. Schlecht (Senior Member, IEEE) received the Diploma in applied mathematics from the University of Trier, Trier, Germany, in 2010, and the M.Sc. degree in digital music processing from the School of Electronic Engineering and Computer Science, Queen Mary University of London, London, U.K., in 2011, and the Doctoral degree with the International Audio Laboratories Erlangen, Erlangen, Germany, on artificial spatial reverberation and reverberation enhancement systems in 2017. He is currently an associate Professor for Signal Processing at the Friedrich-Alexander-Universität Erlangen-Nürnberg, Germany. This position is part of the Chair of Multimedia Communications and Signal Processing. Until 2024, he was Professor of Practice at Acoustics Lab, Aalto University, Department of Information and Communications Engineering, and Media Lab, Department of Art and Media, Aalto University, Espoo, Finland. From 2012 to 2019, he was also an External Research and Development Consultant and Lead Developer of the 3D Reverb algorithm with Fraunhofer IIS, Erlangen.