AES 2025a

The following audio examples accompany the submission

H. Helmholz, J. Crukley, and J. Ahrens, “Perceived Quality of Binaural Rendering From Baffled Microphone Arrays Evaluated Without an Explicit Reference, Part 2,” submitted to AES 158th Convention, Warsaw, Poland: Audio Engineering Society, 2025.

The given binaural renderings provide auralizations for all trials and conditions presented in the perceptual study, according to Table 1 in the manuscript. The auralized acoustic environments (Hall, Lab) and the nominal sound source at the sitting height of a typical adult listener (ear) are shown in Fig. 1. The study conditions comprise array configurations from the types of spherical microphone array (SMA), equatorial microphone array (EMA), and equatorial non-spherical microphone array (XMA). They are presented in study trials with the microphone arrays reproduced at 2nd, 4th, and 8th spherical harmonics order (SH2, SH4, SH8), respectively. Additionally, four conditions (MH comprising captured binaural room impulse responses of the mannequin head and three SMA at SH1, SH4, and a high-accuracy condition HiAc at either SH29 or SH44, depending on the room environment) are presented as repeated anchors in each trial of the employed multi-stimulus category rating (MuSCR) paradigm. These anchor conditions are presented without equalization (EQoff), whereas the non-anchor conditions are presented with equalization (EQon).

The binaural rendering involves captured array room impulse responses and a G.R.A.S KEMAR mannequin head with minor modifications. Equalization filters for the utilized Sennheiser HD650 headphones were employed during the perceptual study, including phase linearization by individual measurement on the KEMAR mannequin head. These equalization filters are also applied in the present listening examples, i.e., the reproduced timbre and externalization may be affected when listening with different headphones. The auralized program materials are short anechoic excerpts of a female speaker (speech) and a percussion stimulus (drums) looped for real-time convolution in the perceptual study.

All listening examples are precomputed for two static head orientations, 0° (frontal) and 90° (lateral), according to the two nominal source directions in Table 1. Furthermore, we provide a precomputed rendition of a continuous horizontal head rotation (0° to 360°). The conditions were presented to the participants with real-time head tracking during the perceptual study (see experimental resources).

The given audio examples may sound very quiet. In this case, increase (or maximize) the output level of your headphones. The normalization is required to prevent clipping in the rendered audio signals while preserving the full dynamic range of the measurement data.

Hall + speech trials (3 array orders × 2 source directions)

Hall + drums trials (3 array orders × 2 source directions)

Lab + speech trials (3 array orders × 2 source directions)

Lab + drums trials (3 array orders × 2 source directions)