My issue with use of the term 3D with reference to binaural voice conferencing service stems from the fact that I’ve been having some fun with real 3D audio over the past couple of months. This has come about since Hector Centeno released AmbiExplorer for Android, an application that lets us decode Ambisonic recordings for binaural playback in headphones.
AmbiExplorer plays back first-order B format files as well as UHJ encoded stereo files. As opposed to simply left, center, right, etc. a B format file is actually comprised of four channels known as W, X, Y & Z. These correspond to three directional signals and an omnidirectional reference signal.
By performing some matrix math on these signal in a coordinated fashion the entirety of the soundstage can be transformed in 3D space. AmbiExplorer decodes the four signals into a binaural feed, allowing the user to rotate their listening perspective in real-time. You can think of it as a 3-axis “balance” control.