Evaluation of Binaural Renderers: Multidimensional Sound Quality Assessment

Reardon G, Genovese A, Zalles G, Flanagan P, Roginska A. Evaluation of Binaural Renderers: Multidimensional Sound Quality Assessment. In: AES Conference: 2018 AES International Conference on Audio for Virtual and Augmented Reality. Audio Engineering Society; 2018. Paper P2-5. Available from: https://aes.org/publications/elibrary-page/?id=19694

@inproceedings{Reardon2018_19694,
  author = {Reardon, Gregory and Genovese, Andrea and Zalles, Gabriel and Flanagan, Patrick and Roginska, Agnieszka},
  title = {{Evaluation of Binaural Renderers: Multidimensional Sound Quality Assessment}},
  booktitle = {AES Conference: 2018 AES International Conference on Audio for Virtual and Augmented Reality},
  note = {Paper P2-5},
  year = {2018},
  month = aug,
  publisher = {Audio Engineering Society},
  url = {https://aes.org/publications/elibrary-page/?id=19694}
}

TY  - CPAPER
TI  - Evaluation of Binaural Renderers: Multidimensional Sound Quality Assessment
AU  - Reardon, Gregory
AU  - Genovese, Andrea
AU  - Zalles, Gabriel
AU  - Flanagan, Patrick
AU  - Roginska, Agnieszka
T2  - AES Conference: 2018 AES International Conference on Audio for Virtual and Augmented Reality
M1  - Paper P2-5
PY  - 2018
DA  - 2018/08/06
UR  - https://aes.org/publications/elibrary-page/?id=19694
PB  - Audio Engineering Society
LA  - en
AB  - A multi-phase subjective experiment evaluating six commercially available binaural audio renderers was carried out. This paper presents the methodology, evaluation criteria, and main ?ndings of the tests that assessed perceived sound quality of the renderers. Subjects appraised a number of speci?c sound quality attributes—timbral balance, clarity, naturalness, spaciousness, and dialogue intelligibility—and ranked, in terms of preference, the renderers for a set of music and movie stimuli presented over headphones. Results indicated that differences between the perceived quality and preference for a renderer are discernible. Binaural renderer performance was also found to be highly content-dependent, with signi?cant interactions between renderers and individual stimuli being found, making it dif?cult to determine an “optimal” renderer for all settings.
ER  -