AES Fall Online Convention

Interior panning algorithms enable content authors to position auditory events not only at the periphery of the loudspeaker configuration, but also within the internal space between the listeners and the loudspeakers. In this study such algorithms are rigorously evaluated, comparing rendered static auditory events at various locations against true physical loudspeaker references. Various algorithmic approaches are subjectively assessed in terms of; Overall, Timbral, and Spatial Quality for three different stimuli, at five different positions and three radii. Results show for static positions that standard Vector Base Amplitude Panning performs equal, or better, than all other interior panning algorithms tested here. Timbral Quality is maintained throughout all distances. Ratings for Spatial Quality vary, with some algorithms performing significantly worse at closer distances. Ratings for Overall Quality reduce moderately with respect to reduced reproduction radius and are predominantly influenced by Timbral Quality.

A method for representing the three-dimensional radiation patterns of instruments/performers within artificial reverberation using multichannel direct sound files convolved with channel-based spatial room impulse responses (SRIRs) is presented. Two reverb conditions are studied in a controlled listening test: a) all SRIR channel positions are convolved with a single monophonic direct sound file, and b) each SRIR channel position is convolved with a unique direct sound file taken from a microphone array surrounding the performer. Participants were asked to adjust the level of each reverberation condition (relative to a fixed direct sound stream) to three perceptual thresholds relating to source- and room- impression. Results of separate three-way within-subject ANOVAs and post-hoc analysis show significant interactions between instrument / room type, and instrument / reverb condition on each of the three thresholds. Most notably, reverb condition b) required less level than condition a) to yield perceptual equivalency between source- and room- impression, suggesting that the inclusion of multichannel direct sound in SRIR convolution may increase the salience of room impression in the immersive reproduction of acoustic music.

This paper describes a comprehensive study on the sound quality of the Opus codec for stereo, surround and immersive audio formats for music and cinematic content. We conducted three listening tests on Opus encoded stereo, 5.1 and 7.1.4 test samples taken from music, cinematic and EBU files encoded at bit rates of 32, 48 and 64 kbps per channel. Preliminary results indicate that a bit rate of 64 kbps per channel or higher is required for stereo, but 48 kbps per channel may be sufficient for surround and immersive audio formats.

We present a subjective evaluation of various 3D main-microphone techniques for three-dimensional binaural music production. Forty-seven subjects participated in the survey, listening on headphones. Results suggest that ESMA-3D, followed by Decca tree with height, work best of the included 3D arrays. However, the dummy head and a stereo AB microphone performed as well if not better than any of the arrays. Though not implemented for this study, our workflow allows the possibility to include individualized HRTF's and head-tracking; their impact will be considered in a future study.

Ambisonic recording with spherical microphone arrays (SMAs) is based on a far-field assumption which determines how microphone signals are encoded into Ambisonic signals. In the presence of a near-field source, low-frequency distance-dependent boosts arise in SMAs in similar nature to proximity effects in far-field equalized directional microphones. In this study, the effects of near-field sources on Ambisonic signals are modelled analytically, their interaction with regularization stages is observed, and then traced further across to two basic ambisonic processing operations: virtual microphones, and binaural decoding.

Although Ambisonic sound reproduction has an extensive history, it started finding more widespread use in the past decade due to the advances in computer hardware that enable real-time encoding and decoding of Ambisonic sound fields, availability of user-friendly software that facilitate the rendering of such sound fields, and recent developments in immersive media technologies, such as AR and VR systems, that prompt new research into spatial audio. In this paper, we discuss the design, implementation, and evaluation of a third-order Ambisonic system in an academic facility that is built to serve a range of functions including instruction, research, and artistic performances. Due to the multi-purpose nature of this space, there are numerous limitations to consider when designing an Ambisonic sound system that can operate efficiently without interfering with the variety of activities regularly carried out in it. We discuss our approach to working around such limitations and evaluating the resulting system. To that end, we present a user study conducted to assess the performance of this system in terms of perceived spatial accuracy. Based on the growing number of such facilities around the world, we believe that the design and evaluation methods presented here can be of use in the implementation of spatial audio systems in similar multi-purpose environments.

In augmented reality (AR) environment, audio signals of real world and virtual world are simultaneously presented to a listener. It is desirable that a virtual sound content and a real sound source do not interfere each other. In order to make it possible, we have examined spatial auditory masking between maskers and maskees, where maskers are real sound signals emitted from loudspeakers, and maskees are virtual sound images, generated by using head related transfer functions (HRTFs), emitted from headphones. Open-ear headphones were used for the experiment, which allow us to listen to the audio content while hearing the environmental sound. The results are very similar to those of the previous experiment [1, 2] where masker and maskee were both real signals emitted from loudspeakers. That is, with a given masker location, masking threshold levels as a function of maskee locations have symmetric property with respect to the frontal plane of a subject. Masking threshold level is, however, lowered than the previous experiment perhaps because of limitation of sound image localization by HRTFs. The results indicate that spatial auditory masking of human hearing occurs with virtually localized sound images in the same way as real sound signals.