Until version 8.0, you technically can but it is not supported (and won't be possible in 8.0). Any prompt files that you want to use are supposed to be in the Prompt Repository.
I must say I'm a little curious what you have designed that requires that many WAV files; however, if a bunch of them are verbiage, you may want to look at using an ASR server to speak the verbiage instead where it makes sense. That would cut down on the volume.