ScatAR_WGW: Implementation and Evaluation of the Waveguide Web in an Application for Artificial Reverberation in a Virtual Environment
Author
Holfelt, Jonas
Term
4. Term
Education
Publication year
2018
Submitted on
2018-05-31
Pages
62
Abstract
Realistisk efterklang er afgørende for overbevisende lyd i virtuel og augmenteret virkelighed (VR/AR). Efterklang, de mange refleksioner af lyd fra overflader, hjælper os med at vurdere et rums størrelse, form og materialer. At simulere disse refleksioner kaldes rumakustisk modellering og kan udføres med forskellige algoritmer. To nyere tilgange er Scattering Delay Networks (SDN) og Waveguide Web (WGW). Dette arbejde implementerer WGW i en eksisterende augmented reality-applikation, der allerede anvender SDN-baseret efterklang. Applikationen bruger de registrerede positioner af lydkilden og lytteren til at indsamle information om omgivelserne og tilpasse effekten. Vi evaluerede begge metoder i et virtuelt miljø. Lyttere oplevede tydelige perceptuelle forskelle og foretrak signifikant SDN frem for WGW. I denne sammenhæng var WGW for beregningstung til realtidsbrug, men rummer lovende muligheder for fremtidigt arbejde, især til modellering af udendørs akustiske rum.
Realistic reverberation is essential for convincing audio in virtual and augmented reality (VR/AR). Reverberation, the many reflections of sound from surfaces, helps us judge a space’s size, shape, and materials. Simulating these reflections, known as room acoustic modeling, can be done with different algorithms. Two recent approaches are Scattering Delay Networks (SDN) and the Waveguide Web (WGW). This work implements WGW inside an existing augmented reality application that already uses SDN-based reverberation. The application uses the tracked positions of the sound source and the listener to gather information about the environment and adapt the effect. We evaluated both methods in a virtual environment. Listeners noticed clear perceptual differences and significantly preferred SDN over WGW. In this setting, WGW was too computationally demanding for real-time use, but it offers promising directions for future work, especially for modeling outdoor acoustic spaces.
[This abstract was generated with the help of AI]
Keywords
Documents