This was very neat, thank you. Dolby Atmos takes the win in my book. Most natural sounding ambiance, no overt errors, I actually preferred the elevated bass resonance. RU-vid and Facebook feel surprisingly tied for second. Auro had so much unnecessary reverb that it was distracting to me, and I can't imagine that it's not a bug. MPEG-H failed to make the whisper at 16:56 sound like it was actually right in my ear, so it is instantly disqualified, even though I appreciate its ability to separate dialog from the broader mix, for the hearing-impaired. If it means anything, my 'test bed' was some HIFIMAN HE-4XX open-back planar headphones with a 1 watt USB amplifier; I did notice needing to crank the volume practically to max for the demo, though.
That last demo was good. As you said, the whisper with MPEG-H wasnt where it was supposed to be (right in your ear). I actually preferred how Auro3D did it best. That reverb is supposed to be Auro's thing. Its mimicking reflections in the environment, as a reault it sounds a lot more spacious and enveloping. I like that effect. Atmos sounded good....the tones were natural sounding with a good sense of space and directionality, but not as spacious sounding as Auro3D. Where Atmos dropped the ball was with the "gentle stream" and distant thunder. I could barely hear it at all. The stream sounded more like a babbling brooke and the distant thunder sounded like wind hitting a microphone 😆 Facebook 360 sounded best with that part with Auro3D right behind it and RU-vid in 3rd place.
Would have loved to see how the IEM Suite, Sparta Suite, 3DX, Google Resonance Binaural Decoder, Blue Ripple Suite or Mach 1 System compares to those...
This video was mostly focused on consumer delivery formats, but in the future I may look at some of the other binauralization options like the ones you listed!
I always enjoy your videos. And I have to give you props for the bow-tie; that's a good look. For me, anyway, perhaps because I've not heard good implementations of it, it hasn't been truly 3D convincing. Given my limited understanding, the pinae shape used to house the two microphones used for capture would need to be a fairly close match to the listener's pinae shape, or the simulation modeling for output would need to be tailored to match the listener's brain, for it to be reasonably convincing (but you're far more knowledgeable, and I could be completely wrong).
There's studies showing our brain can adapt to an HRTF model that doesn't perfectly match the ear, but it takes time and a visual reference to match to. It would make sense that since I've been listening to all these formats, my brain has already at least slightly adapted to them, but I totally get being in the situation where you can't hear much of a 3D effect at all. Sometimes it just doesn't work.
Hey man! I like how you chose to use a field recording with a substantial noise floor as a base. That is where i hear the clearest difference. Which is pretty significant. Is this due to HRTF´s? There is also a difference in reverb. I especially heard it on Auro 3D on the sine tone thing. Alot more audible reverb on that compared to the other.
Any differences in tone/frequency are due to the HRTF. Several of the models have some amount of reverb, but Auro was definitely the most dramatic with it.
The file size would depend on how exactly Facebook vs RU-vid are encoding their files, but if you're using the same compression method, 2nd order requires about twice as much data.