How it works
Creation of sound geometry scan and audio signal processing takes up to 10% of CPU computational capabilities, using modern commodity hardware, which is a standard nowadays (http://www.maximumpc.com/everything_you_wanted_know_about_amd%E2%80%99s_...). By "modern CPUs" we mean architectures that support AVX instruction set (introduced in 2011), namely last 3 gens of processors starting with Intel's Sandy Bridge and AMDs Bulldozer cores. Keep in mind, that sound path creation algorithm supports phenomena like specular reflections, diffuse reflections and edge diffraction, whilst audio processing supports signal filtering (modeling air dumping, reflection/diffraction impact and so on), Doppler effect modeling and spatial effects. In the presented video directional sound is implemented using binaural auralization per each sound path.