You can quickly and easily audition "Voice" Libraries without having to load the Voices to internal User bank location. This is done with the "Motif XF Editor Standalone/VST" and the XF's Edit Buffer (as explained in this RESOURCES article: Auditioning Voices from Voice Libraries

However, if you have a Sample Library (a Voice Library that also includes new Samples and Waveform) then, obviously some additional transferring is necessary. You cannot stream 128 audio channels from a simple USB drive. This is why you cannot audition Voices that contain samples from your USB stick. You must transfer that sampled audio data to the XF's RAM (or to a FLASH Board) which is a way, way, way faster type of memory (and a lot more expensive) than any USB storage type device's memory.

Currently USB drives can stream stereo audio, (with a fair amount of latency), ...the SDRAM of your XF, on the other hand, is fast enough to retrieve your audio in nano seconds, and can stream audio via 128 audio channels (128 notes of polyphony) - so you will need to load the Waveform and Samples to the XF in order to audition Voices that come from Sample (Voice) Libraries. Sorry, but there is no way around that.

Normal Voice Libraries can be auditioned by loading the Voice parameters to the Editor and have the Editor "bulk" the data to the XF's edit buffer. This works because the Voice data points to a known location (preset) for the Waveform... In normal Voice Libraries the location of the Preset Waveform 0001~3977 is known.

When a Library contains new samples, these are always installed to the XF's memory "dynamically" - that is they find the lowest numbered empty location, and the Voice data is automatically updated by the XF hardware.

What we recommend
Before auditioning your Sample Library, make an ALL data file backup of your current situation (this way you can return to it)
Set the TYPE = 1 Bank Voice
Highlight the Sample Library ALL data file
Press [ENTER] to open the File
Select the Folder containing the Sample Library's Voice data
Direct it to the internal Voice bank in your XF - I recommend "User 1"

Set the FLASH Load Options (lower right of the File screen):
USR > USR
FL1 > USR
FL2 > None

By redirecting FL1 to USR when you execute the Load, the waveforms and samples will occupy your User SDRAM... The XF hardware will dynamically update each Voice so that it properly finds its data.

The Voice parameters will, indeed, occupy your USER 1 bank, having overwritten the previous data. You can audition the Voices as normal.
I recommend you make notes of those Voices you desire to keep. Later you Load them individually, pointing the Load Option "FL1 > FL1" which will install the keeper's data to your Flash Board. And as per usual: after installing anything new to a Flash Board you need to update your Archive Backup File (ALL "with samples").

After you have finished auditioning the Library you can restore the data you previously had in USER 1 by again using the "1Bank Voice" option from your backup file... (Or you can restore the Factory User 1 Bank by simply going to UTILITY > JOB > set the function to restore just USER1)

Hope that helps.