Impressive as the new Emsemble feature is on the T5, the horns don't sound realistic. Yamaha has come a long way with the SA2 slice & splice technology, but it can never achieve the articulation of an actual horn or flute player... unless the Polyphonic Aftertouch can be mapped to Velocity or Volume... which would require a new playing technique that may not be so easy for some.
This reminds me of the VL technology that came out in the late 1990s. The VL-1 and follow-up sound modules required a whole new way of playing using a Breath Controller or WX Wind Controller learning to tweak the VL parameters to your personal playing style. It looks like learning to play the Ensemble Voices with 1,2,3,or 4, Voices requires concentration and a new playing technique.
This is a wonderful example of Yamaha's inovations. Unfortunately sometimes these are not embraced wholeheartedly by the consumer because of the difficulty in learning the technology or techniques required to play them to their fullest potential.
As with any new feature, there is a certain learning curve to maximize the full usefulness such as the Tyros 5 Ensemble feature. The realism or unique concept was to simulate the ensemble of various instrument combinations controlled automatically from your keyboard playing, allowing up to four different instruments to sound at a time ( playing manually or through style auto harmony ).
More of the individual or solo instrument voices are better served through the SA2 voices. The ensemble voice combinations are going to require some playing technique or a playing technique that would simulate the instruments and the interaction in a real ensemble. Yamaha did a pretty good job with the ensemble voice presets. You can easily edit the presets to your liking or playing style. The voices will not work like SA2 voices, but other ensemble and harmony parameters are available.
As with my Tyros 5 styles, I basically tweaked every style to my liking and in the same manner tweaked the ensemble voices for my own edit saves. The only additional parameter I wished was there is aftertouch. Aftertouch control could of been more handy for more user expression, but I think the concept leans more towards the instruments working together as an ensemble, not trying to control expression for each individual instrument voice.
If you select ensemble presets that are close to what you want, you can go into edit and adjust the various parameters or change to entirely different individual voices. Probably the most useful parameter change is the key assign type. A more suitable key assign type might help lean towards your playing style or technique. It took me quite a bit of experimentation to get the settings I liked.
I saved my ensemble voices by category folders on my hard drive and created registration banks to access the voices that way. Once saved in this manner, I was also able to save the ensemble combinations to the 4 OTS slots for certain styles. The ensemble feature will still work in the exact same manner even through OTS saves. Simply awesome for live performances once all your saves are made.
Coincidentally, I still have my Yamaha MU100r sound module that came with a VH board and the VL board you spoke of. I did get interesting effects and sounds with a wind controller and changing the various pitch, filter, amplitude, and LFO depth with this physical modelling technique, it took a lot of practice and tweaking to simulate a realistic instrument.
Seeing all these innovations through the years, the most realism I ever seen or heard is through the SA2 technology for individual instrument voices, and perhaps the Tyros 5 ensemble feature to control ensemble voicing. All these technologies and techniques require a certain amount of practice so that the user expression can ultimately show through. I hope this helped somewhat.