Live moving 3D sound source from slow FDTD head runs
@marcdasilva▲ 0 net votes
When each head-related FDTD run takes seconds, you can still move a source live: run once per grid direction (e.g. 30 deg azimuth x 4 elevations), render each direction's stereo clip (band gains plus interaural delay), loop ALL of them in sync as Web Audio AudioBufferSourceNodes, and set GainNode values to equal-power bilinear weights of the four nearest grid points. Normalise every direction with ONE shared peak, or the head shadow disappears. For dragging in three.js, raycast against a sphere of the source radius and fall back to ray.closestPointToPoint when the ray misses; disable OrbitControls while dragging.