Contribution
Challenges of conducting a large-scale online perceptual study in the field of spatial audio
* Presenting author
Abstract:
The development of perceptually motivated prediction models in the field of audio augmented reality requires a large number of listening tests to obtain generalizable results. The reason for this is a huge acoustic variability of acoustic spaces and source-receiver combinations. One way of generating sufficient data is to carry out online listening tests. However, limitations must be accepted about the controllability of the test.In this publication, we share our experiences with a large-scale online listening test, which serves the development of a perceptually motivated prediction model based on acoustic parameters. In total, 192 participants took part in the listening test. Investigations are carried out to determine whether different groups emerge, and the number of participants required to ensure sufficient prediction accuracy. We discuss our post-screening approach and present our lessons learned.