The helpful answer has already been given by me way up the thread. Quote " Yamaha imply natural means extra long samples, or maybe more samples per voice." If natural have extra long samples, then it follows that normal must have normal samples i.e. less memory allocated. Particularly if natural are specifically promoted for high end Clavinova in the numbers quoted in the first post, and the non-cool, non-sweet, non-live, non-mega voices on 9000 and tyros are not promoted as natural but specifically called normal instead by Yamaha themselves.

As to bias on my part, none of my statements show any factual errors, where is the bias? In contrast you persist in assuming natural is the same as normal when Yamaha say there are specific numbers of natural voices on high end Clavinova but no natural voices on 9000 or tyros. Then you tell us that George's post "does not help to clear that up" when George specifically states "Natural voices are not used in TYROS"...