I haven't had time to download and listen but there is a known issue (with all types of music processing that use FFTs/Phase Vocoders) with noisy signals like snare drums and cymbals, etc. you get a "swishy" sound. The solution (sorry) is up sample your song first, run through SpecWeb, then downsample. E.g. if your song is 44.1KHZ upsample (audiomuxer does a great job of this) to 88.2KHz 24 bits, run through SpecWeb, and then downsample the result back to 44.1 (but keep the 24 bits).
If I could figure out how to "oversample" in SpecWeb I would ;0)
Anyway, this is probably one difference between what folks are doing here and the "DKA" etc. conversions with the full version of Spec, most folks using the full version always up/down sample.
However, don't do it just 'cause I said. Do it if it makes your results better ;0)