Loading…
Schedule as of Oct 11, 2022 - subject to change

Default Time Zone is EDT - Eastern Daylight Time


Back To Schedule
Thursday, October 20 • 2:00pm - 2:20pm
Algorithmic Methods for Calibrating Material Absorption Within Geometric Acoustic Modeling

Sign up or log in to save this to your schedule, view media, leave feedback and see who's attending!

In room acoustic modeling, geometric acoustic (GA) room models are commonly created to aid acousticians. It is critically important to have the parameters of the room and simulation match in order to assure physical realism within the model. Traditionally, acousticians manually adjust absorption coefficients of planes in the model to align simulated parameters, such as reverberation time (T30) or speech clarity (C50), with values measured from the real room. One of the largest setbacks to this method of manual calibration is the time consuming process of adjusting the acoustic coefficients of planes in the model. The acoustician has to ensure absorption coefficients stay within a reasonable range while simultaneously deciding which planes to adjust depending on their size and proximity to relevant source and receiver positions. Previously, genetic algorithms have been implemented to calibrate GA models based on acoustical measurements. The largest drawback of the genetic algorithm method is that it does not allow acousticians to have as much control over the calibration procedure. One solution to this problem is to implement a traditional auto-calibration algorithm where acoustic coefficients are adjusted linearly until a JND threshold is reached. In order to prevent it from generating unrealistic results, natural ranges for material acoustic coefficients must also be implemented. Instead of the acoustician having to determine the bounds for the algorithm, this express paper proposes that a dataset with measured variance should be used as algorithm bounds. The database illustrated in the express paper can be further developed to be used as a calibration reference for other heuristic and meta-heuristic algorithms. This would allow absorption coefficient deviation to be empirically derived as opposed to being estimated by the acoustician.

Speakers
avatar for Noah Deetz

Noah Deetz

American University
Noah Deetz is currently a Junior attending American University's B.S. in Audio Technology program in Washington, DC. Most recently, Noah completed a research fellowship in acoustics under the NASA DC Space Grant Consortium. 
avatar for Braxton Boren

Braxton Boren

American University


Thursday October 20, 2022 2:00pm - 2:20pm EDT
2D04/05