The main reason why it would fail is the fact that audio is typically sent transcoded, resampled and/or processed through multiple software stacks. If you can get the VMWare Horizon to send the audio in a lossless manner and set both the sender and recipient devices to the same recording rate (e.g. 16-bit 44.1khz) you might get it to work.
I've had mixed success with using Proximity in scenarios where the audio passes through multiple layers of software abstraction (e.g. virtualization, remote desktop.) even when I've tried to follow that guideline.