I'm quoting from the FAQ:
"Does Groovyband Live! produce the sound as well as the realtime arrangement?
No. Groovyband Live! produces only a stream of midi data. You need a compatible sound generator to actually produce sound."
This means, a saxophone sample of an older PSR arranger will always sound like exactly that sample of the older PSR arranger, and never like the newer sample of a Genos, nor will it have the SA2 features of automatic slides etc.
While it is certainly true that you cannot have a SA(2) voice on PSR-S(X) model, the reality is not as bad as it might seem at first sight.
1) SA voices are mainly useful for right hand parts. For style parts (which are preprogrammed) all (or most) the realtime tricks built-in a SA voices can in fact be already applied in the style pattern.
Yamaha itself does NOT use SA voices for (most ?: we did not check them all!) style parts. Hence the automatic style arrangement is totally (or largely) insensitive to the presence of SA voices.
2) Having "more" voices is subject to the law of diminishing returns.
Having 2 guitar voices instead of 1, is a 100% improvement. Having 25 guitar voices instead of 22 is only a ~9% improvement. When you apply a stomp box effect to those guitar voices then you suddenly do NOT recognize anymore the original sample. And having 25 instead of 22 (or 5!) is largely meaningless.
3) Most top-hits rock bands use only 1 (or 2) guitars for the whole concert. Nobody in the whole story of rock ever complained for a lack of guitars!!
4) The same can be said for the number of effect algorithms available.
For example, even the most humble Yamaha arranger these days has 4÷5 different reverb algorithms (we do not have here the data list to check the exact number). Can you tell by listening to a playing style which is which? Do you feel the need to have one more to improve the sound?
Out of 90 or so different DSP algorithms available in a PSR, those used regularly in Yamaha preset styles are 10÷20. Most of those 90 are NEVER used. Do you really think that having 110 instead of 90 will make any practical difference?
5) DSP allocation in Yamaha arrangers is rigid. For example in a PSR you can have only 1 insert DSP for style parts. GBL can freely allocate DSPs where they are most needed. Thus you can do more with less HW resources, and suddenly even a PSR which has half the DSP power of a Tyros can compete roughly at the same level.
Here too the law of diminishing returns applies: being able to apply 2 or 3 DSP to style parts (vs 1 of stock firmware) is a huge improvement. More than that, although nice, is less noticeable, because many instruments do NOT desperately need a DSP as a guitar part calls for. And guess what? Not even Tyros/Genos arrangers use all the DSPs they have on paper! After all, what DSP could sensibly be applied to a drum part for example? Maybe compression, but the effect is subtle and most people do not even notice!
For the last 20 years Yamaha always increased the sample rom giving you more and better sounds. But at a certain point the improvements are only marginal: how better can be a (let's say) sampled piano at every generation? If you want to have fun listen to the product demonstrations of Yamaha's products of the past. Each time they wanted to tell you how great was that damned piano sound. They ran out of adjectives long ago. If it was so good 10 years ago (according to Yamaha own words) how can it be significantly better now? Did they lie in the past or are lying now? And if they lied in the past how can you trust them now?
What they NEVER did was to improve the style playing engine: it is exactly the same as it was 20 years ago. When you start from a modest level, every gain is significant. But they insist in giving you more sample rom (and more DSPs) and nothing else. They insist on improving once again that damned piano!!
The reason is simple: thanks to silicon improvements, having more flash ROM and computing power comes free. Look at what CPU or SSD you can buy now with 200€, and what you could buy for the same price 5 or 10 years ago.
So you almost for no additional costs (sampling sessions are one off costs, recycled/amortized among all the product ranges and countless units sold for decades) and zero risks (business as usual) can dish out a new HW model with more sounds and DSPs. No significant R&D, no creativity, no new ideas.
This works because in the arranger market there are very few (used to be 4, now are 3) players.
* * *
What makes a HUGE difference is the flexibility you have in programming a style. Being able to use at its highest potential what you have, will often compensate for the inferior HW resources on paper. And then some!
A man is not the strongest, nor the fastest, nor the one with the better eyesight, or hearing of the animals. He is not able to fly, he swims poorly.
Nonetheless he was the most successful. Because he was the most clever in doing things.