As it stands now, when preload_rsc is used the client is forced to download the entire package, sounds and all; even if the client has their sounds disabled.
Without preload_rsc the client only has to download the resources they need, cutting out needlessly large downloads for files they aren't going to use.
I understand that using preload_rsc makes it pretty much impossible to determine which files not to download since it's just downloading the whole package.
What I propose is a second variable that works exactly like preload_rsc except allows for a second "lighter" file to be downloaded if the client has their sounds disabled. Which would mean of course providing an option in the package wizard to generate a file without the sounds as well as one with the sounds.
A lot of people, including me leave their sounds off not because the sounds annoy them but because their connection can't really handle downloading 10MB of music they'll never listen to.