AES Fall Convention 2022

Sound sources have complex three-dimensional radiation characteristics that vary over time. In real-world listening, these dynamic fluctuations in directivity map to the reflective environment to which the source is coupled. Artificial reverberation tools for audio post-production do not currently account for this in their designs. A listening test was conducted to compare two artificial reverberation treatments for spatial room impulse response-based convolution over headphones. In the first treatment, all SRIR channels were convolved with a single monophonic direct sound file, recorded with a microphone placed in front of the performer. In the second, each of the 24 SRIR channels were convolved with a corresponding hemi-anechoic microphone position. To quantify the perceptual differences between the two treatments, participants’ sense of physical presence, sensory immersion, environment / room width, and naturalness of the sound source were evaluated. Results of three-way within-subject ANOVAs show that the multichannel direct sound convolution treatment significantly improved participants ratings of all attributes under test, and was preferred over the monophonic direct sound treatment. A significant interaction between reverberation treatment and instrument type was observed for environment / room width. The study suggests that this technique can be effective, and more broadly supports the notion that representing a sound-sources’ time-variant directivity in artificial reverberation could be a beneficial feature in the design of 3D audio production tools.