|
|
|
|
|
Data Driven Acoustic Design , ETH Zurich, 2018-2022
|
PhD research
|
This research aims to develop a novel approach to performance-driven acoustic design of sound diffusive surfaces. This approach will enable designers to explore and design a plethora of acoustically-informed surfaces without requiring expert knowledge in acoustics.
It focuses on collecting, analysing, and classifying impulse responses from computationally designed and physically prototyped surfaces to build a training set for machine learning applications. A state-of-the-art automated robotic setup was used to create the GIR Dataset, an extensive collection of real impulse responses and three-dimensional diffusive surfaces. Unsupervised machine learning techniques and custom data visualisation methods are used to analyse and explore the GIR Dataset. The outcome of this research aims to simplify the design-simulation-evaluation process, bringing acoustics closer to the architecture practice and enabling more acoustic aware designs.
Publications
GIR Dataset: A Geometry and Real Impulse Response Dataset
Data-Driven Acoustic Design of Diffuse Surfaces Using Self-Organizing Maps
Visualization methods for big and high-dimensional acoustic data
Computational Design and Evaluation of Acoustic Diffusion Panels for the Immersive Design Lab
A Data Acquisition Setup for Data Driven Acoustic Design
|
|
|
|
|
Credits:
|
Gramazio Kohler Research, ETH Zurich
Achilleas Xydis, Dr. Romana Rust, Gonzalo Casas, Dr. Beverly Ann Lytle
Laboratory for Acoustics / Noise Control Empa
Kurt Eggenschwiler, Dr. Kurt Heutschi
Strauss Electroacoustic GmbH
Jürgen Strauss
Swiss Data Science Center (SDSC)
Dr. Fernando Perez-Cruz, Dr. Nathanaël Perraudin
Support: Michael Lyrenmann, Philippe Fleischmann (Robotic Fabrication Lab, ETH Zurich)
|
|
De |