It's about an improvement to the ADH technology. In a nutshell, the idea is to reduce the stress on the class A amplifier, particularly at higher audio frequencies, so that it has to deliver less current and therefore is more accurate. This is done by adding a carefully designed loading network to the output of the class D amplifier.
Here's my attempt at translating the patent into something resembling normal English...with apologies for any misunderstandings on my part. (By the way I think there may be several typos that escaped the proof-reader, which threw me a little on first reading.)
Figure 1 shows the class A amplifier (in the box marked 18) and the class D amplifier (19). Their outputs are coupled to each other and the loudspeaker at point 14. This is the ADH core of all Devialet amplifiers. The class A amplifier is a very linear (accurate) voltage amplifier and the class D amplifier provides current to beef-up the output of the class A so it can drive real-world loads like loudspeakers. The class D amplifier inherently produces an on-off, pulse-width modulated (PWM) output which is merged with the class A's inherently analog output via an inductor (32A/B in the figure).
The class D amplifier is switching at a high frequency - effectively 2 MHz in Devialet's design - which is an engineering compromise between two opposing factors. First, at higher switching frequencies the class D amplifier is less efficient, i.e. generates more waste heat, because the transistors that implement the switches take a fixed time to turn on and off, during which they're dissipating lots of power (I x V). To make the class D as efficient as possible, then, you don't want it to switch at too high a frequency. On the other hand, at lower switching frequencies there is a higher ripple current through the inductor that couples the class D and class A amplifiers. The class A amplifier has to absorb that ripple current, which stresses it, causes it to dissipate more heat, and makes it less accurate.
Interestingly as a side note there is a hint (in paras 0016-0018) that operating the ADH system in a bridged, configuration is one way to reduce this ripple current and thereby allow the class A amplifier to be more accurate. Maybe that is a reason why the dual mono set up has a surprisingly large effect on sound quality, more than the increased power capability would suggest?
What the patent covers specifically is the addition of a damped RLC network (the components labelled 44x) that is carefully designed to reduce the ripple current at the class D switching frequency while having no noticeable effect at audio frequencies. This new network is presumably in the Pro amplifiers, while the original D-Premier and Expert series just have a resistor, according to para 0020.
As a result of this new RLC network, the class A amplifier has to "deal with" much lower ripple current from the class D, which makes it both more accurate and less power-hungry. I think this explains why Devialet said the Pro series sounded better because the current delivered by the class A amplifier was reduced relative to the class D.