Exact Convolution with a CV indexed slice of a IR

Just a thought but I came across some old HRTF’s from college days - they are from a public database from MIT on Kemar dummy head measurements. Used for simulating binaural located sources. Remember them being quite decent ( from an out of head perspective

Each is 512 samples long at 48k and are catalogued at every 5 degrees elevation and azimuth. I cascaded 72 files at all (360deg/5) azimuths of a single elevation (0deg/horizon for starters) for each ear. total file size for each channel, left and right is 72x512=36864 samples

Is there anyway to assign such a file, with 72 slices defined (external definition in the lua file?) so that the convolution can be determined by CV?

Would be amazing to pan externalized spatial sounds like this! Maybe it is quite efficient as convolution only 512samples. From experience, to obtain a natural effect, size of the HRIR (being truncated so much) is not so important- follow the simulation however with decent reverb is

Is this a big ask for this Unit?

A more sensible way to go is perhaps externally (matlab/max etc) convolve say a dozen random phase saws each with different spatial filters- would be quite something

1 Like

Touching on CGI for sound,
Yet another rabbit hole, http://soundparticles.com/whatis.html