Categories
Random

Progress in audio localization

So years ago in grad school at UNM, I worked on audio localization. My work was on the idea of doing the convolution (applying the filter to the audio) in the wavelet domain, to extend the results from a phd student in the group. My results were poor, read about them on this page.

Then this week I saw this on Gear Patrol:

Via Gear Patrol

The reviews are mixed, but the really impressive part is less obvious – you use your phone camera and it uses photogrammetry to derive the geometry of your head and pinnae, and from that it creates a personal HRTF that it loads into the finger-sized gadget:

Image from https://us.sxfi.com

Damn. Deriving an HRTF from just a couple of pictures? As opposed to anechoic chambers and KEMAR heads?

NIST via howstuffworks

I’m sorely tempted to buy one just to see how well it works. At $160 is a bit much for an impulse buy though.

It’s really cool to see similar work, with a big chunk of progress plus some really new and clever ideas added. This version is USB, so Android/desktop, but they do have a Bluetooth version in the works that’d work with my phone.