Perceptual Evaluation of the Open Binaural Renderer

Sochaczewska K, Rudzki T, Kearney G, Skoglund J, Lim FSC, Huoponen J. Perceptual Evaluation of the Open Binaural Renderer. In: Convention Paper. Audio Engineering Society; 2026.pp. 10272:1-10. Available from: https://aes.org/publications/elibrary-page/?id=23213

@inproceedings{Sochaczewska2026_23213,
  author = {Sochaczewska, Katarzyna and Rudzki, Tomasz and Kearney, Gavin and Skoglund, Jan and Lim, Felicia S. C. and Huoponen, Jani},
  title = {{Perceptual Evaluation of the Open Binaural Renderer}},
  pages = {10272:1-10},
  url = {https://aes.org/publications/elibrary-page/?id=23213},
  year = {2026},
  month = may,
  publisher = {Audio Engineering Society},
  address = {Copenhagen Denmark}
}

TY  - CPAPER
TI  - Perceptual Evaluation of the Open Binaural Renderer
AU  - Sochaczewska, Katarzyna
AU  - Rudzki, Tomasz
AU  - Kearney, Gavin
AU  - Skoglund, Jan
AU  - Lim, Felicia S. C.
AU  - Huoponen, Jani
SP  - 1
EP  - 10
M1  - Paper 10272
PY  - 2026
DA  - 2026/05/28
UR  - https://aes.org/publications/elibrary-page/?id=23213
PB  - Audio Engineering Society
CY  - Copenhagen Denmark
LA  - en
AB  - This paper presents the perceptual evaluation of the Open Binaural Renderer (OBR), an open-source library developed for headphone-based rendering of Immersive Audio Model and Formats (IAMF) content. The evaluation followed an iterative framework in which findings from a pilot listening study informed the tuning of rendering profiles, and the resulting renderer was benchmarked against established proprietary solutions. In the pilot study, 19 expert listeners rated the Overall Listening Experience (OLE) of the initial prototype (OBRv1) and five external renderers across diverse audio content. Qualitative feedback was analysed using inductive coding to identify salient perceptual dimensions. The pilot revealed content-dependent performance and showed that a single default profile was inadequate, yielding mixed responses in both the numerical scale and in the qualitative feedback and motivating the development of multiple rendering profiles in OBRv2. The main study evaluated two OBRv2 profiles targeting different reverberation characteristics (Direct and Ambient) alongside three top-performing external renderers. A total of 39 participants, divided into expert and non-expert groups, rated five perceptual attributes: Voice Quality, Envelopment, Externalisation, Overall Listening Experience, and Timbral Balance. Mixed-design ANOVA revealed significant main effects of renderer condition on all attributes. Pairwise comparisons showed that OBRv2;Ambient achieved significantly higher OLE ratings than one proprietary renderer and reached statistical parity with the remaining two, representing a measurable improvement over the prototype. A trade-off between Voice Quality and Externalisation was observed, driven by the level of reverberation in each renderer. The results demonstrate that iterative, perceptually informed tuning can yield competitive binaural rendering quality in an open-source framework.
ER  -