Since completing and tweaking the Lite LS-29 (clone of the X10), though it would be interesting to compare the stock X10D vs the tweaked LS-29 @ http://myoldvintagehifi.blogspot.com/2010/11/diy-tube-buffer.html
The stock X10D uses Philips 6922 valves and produces a rich mid, nice high with a lean low after power-on. About 15-30min later, the bass becames fuller but I find the high can be just a tad brighter than an actual presentation should be. The X10D seem to have a slight roughness around the edge as well, particularly on voices. I was using a gold CD recording of a singer whom I heard during a live performance for comparison.
The LS-29 produces a more authoritative low and "richer" high while retaining the rich mid of the X10D. The reproduction of the LS-29 was more natural.
Finally decided to recap using lessons learnt from the LS-29.
Upon power on there was a immediate bandwidth improvement, compared to the stock X10D which was more reserve in presentation. Voices were now more natural and realistic.
After a week or so, decided to change the 0.22uF Wima(s) to the older MKS-C 0.33uF in the audio section. Initially sounded overly bassey. After a couple of weeks, it's sounding very close to my Tubelog when performing A-B comparisons - with a bit more HF extension, maybe due to the use of silver mica's in th output stage.
Update - 18Aug2011
Finally swapped the Philips 6922 with the Russian 6N23P-EV. Reproduction clarity improved.