So the Tyros allows the player to extend the "body" as long as desired and then change to the "tail" with a sampled vibrato? That's what should be implemented in the Roland SN tones. But maybe the difference is (that was recently discussed in a German forum) SA2 voices are still sample-based and Roland SN tones may be more real physical modeling. But in the latter case, the modeling for vibrato would have to be made more realistic, but then it would demand more processor power, as the Integra already struggles with several SN tones played simultaneously. So I guess the "simpler" scripting/sample approach of SA2 voices is the easier and better solution. Quite a complex matter, this modeling...