View Single Post
Old 24th August 2019
Lives for gear

I've looked into this quite a bit, and I think at least some of the lack in spatial resolution and inability to distinguish between front / back and height results from a mis-match in the HRTF used to convert the ambisonic stream to binaural. You can download and test different HRTF decoders from Blue Ripple, Google, Facebook, IEM, Sparta, etc., and you'll notice they all sound different - sometimes drastically different, again depending on how closely each aligns with the shape of your head, ears, and torso - i.e. your custom HRTF. The only way around this right now is to either use a speaker rig that doesn't use HRTFs, or to get a custom HRTF scan, then load the resulting SOFA file into an ambisonic to binaural decoder that allows you to use your custom HRTF. I have not gotten a custom HRTF yet, but to my ears, the Blue Ripple "headphone" decoder in one of their paid packages sounds the best to me, both in terms of spatial resolution and overall sound quality. I don't think the lack in quality has anything to do with the Facebook 360 Spatial Workstation per se, but rather in the nature of ambisonic to binaural decoding. Also, you'll get a bit better spatial resolution if you up-sample first order recordings with either Harpex or Blue Ripple's harpex Upsampler, but it's not magic. You'd also have better overall quality using a 2nd order mic like the Core Sound Octomic, which benefits from the higher native order recording plus a custom filter for each mic that results in a better a to b-format conversion than a stock filter for an entire line of microphones, like the zoom / ambeo uses.