Fine Tuning of Aperiodic Ordered Structures for Speech Intelligibility

Speech intelligibility is crucial in many spaces, yet designers often fail to predict the acoustic shortcomings of certain design choices. This paper builds on the potential of hybrid surface treatments showcasing low-frequency absorption to control background noise levels and high-frequency diffusion to improve speech-in-noise perception to introduce a workflow that encodes this information in a format easily perceived by designers. After patterns are being classified based on periodicity into partly periodic, non-periodic or aperiodic, a matrix serves as a rule of thumb communicating to non-experts the critical variables for high-frequency diffusion, such as well depth sequence, scale and profile. These become inputs of a computational process that generates variations to tailor patterns for speech
intelligibility. Lastly, plotted graphs that visualize quantitative figures obtained from simulations are marked by a bounding box relative to the effective frequency range for designers to evaluate examined patterns during the process of optioneering. This integrated workflow targets architects and designers that seek for visual feedback to support an iterative exploration of performance driven geometries, while recognizing the contribution of aperiodic order to uniformly distribute the flow of sound energy.