As a Roblox developer, it is currently too hard to create an immersive experience with audio. As I am currently working with a dev team on a project, I will focus on one part of the issues with the Roblox sound system. Our sound designer will add more specific requests to this topic.
In our current project, sounds need to be precise to the millisecond. Any time gaps or deviations from this level of precision will kill the immersion.
There is a few ms delay from when a looped sound ends to when it restarts its loop. This creates a subtle yet noticeable gap in, for example, character movement sound loops. This gap becomes even more detrimental to the experience when the sound loop corresponds to an animation loop, as the sound and animation tracks fall out of sync with time.
It is technically possible to resolve this by manually controlling the loop via code. However, there is no reason why this should be necessary. Unity and Unreal Engine seamlessly loop audio tracks (using Wwise), and Roblox’s sound system would be significantly improved by doing the same.