From the first post of this thread:

* It connects via USB to the synth and drives its sound engine via midi.
* 1800+ custom developed AWM2 and FMX sounds optimized for arranger use. Massive use of available DSPs.


To elaborate further, from our website:

THE BASELINE
We started from the factory sounds and quickly realized that they are not suitable for the purpose:

* Unequal volumes and character, no standard style of programming. They are a collection of sounds gathered together from different sources and legacy synths with no relation with each other.
* Wildly inconsistent response to controllers.
* Multi parts occupied (even when not necessary).
* Unpredictable use of DSPs.
* Lack of sounds/choices in many categories important for arranger use.
* Ridded with arpeggiated dance-style riffs which get in the way while live playing in most musical genres.
* Mostly inherited from 20+ years old synths, with sparse or no use of the newest and most exciting waveforms.

WHAT WE IMPLEMENTED
So we decided to reprogram the whole set from scratch in a consistent and predictable way:

* All the sounds have the same volume (you can swap them arbitrarily without the need for any volume correction in the mixer).
* All the sounds come with a variation built-in, thus instantly doubling your choices. The variation is sometimes bold, sometimes subtle, but always useful and appropriate for the sound at hand.
* All the sounds come with a main modulation (usually acting on many synth engine's parameters), tied to the mod wheel. You have one easily/naturally reachable and consistent point of modulation during your performance.
* All the sounds are editable with a handful selection of top level parameters, those that make more sense for an arranger player. You obtain the most impact with the least cognitive effort.
* All the sounds are single part (= 1 midi channel). You have 16 parts in the mixer, and you can use 16 independent sounds. Simple, effective and well tested paradigm; universally in use since the inception of midi music.
* All the sounds make consistent use of DSPs: by design they sound good even without them, but can optionally be enhanced/changed with them. The preprogrammed DSPs do NOT alter the overall volume of the sound (only the character), so you can switch them on/off seamlessly. Consistency and interchangeability are milestones of our philosophy.
* We added many patches, both AWM2 and FMX, filling the gaps and increasing the diversity of the available sound palette. We also programmed many "mega" voices, used by the arranger engine to increase the realism of acoustic sounds.

BENEFITS FOR THE PLAYER
The result of this monumental effort is a gigantic set of 1800+ high quality and fully consistent sounds (AWM2 and FMX). For the first time you can use native FM voices in an arranger.

Never such a comprehensive sound set, designed to play like a team in a predictable and consistent way, has been available in a sound module/synth.

All the sounds are thought to be played by hand and are useful in real live performances by real players (not a single sound makes use of arpeggiated stuff). You can creatively use Groovyband Live! even without touching the arranger engine: use it like a workstation/sound module with tons of quality sounds you can easily organize in split/layers and play from up to 3 keyboards. Everything from a clear, touch native, user interface which conforms to well tested industry standard paradigms everybody is used to.

To make things more exciting, you can use all these 1800+ sounds without committing a single memory location in your MODX/Montage: we program on the fly the performance buffer (a volatile memory location meant to hold the "currently playing" performance). Groovyband Live! thus acts in a fully transparent manner: it does not modify your synth in any why, and leaves it exactly as it was found!


If you check the website there are many more info and also the full user manual.