Oooh "technically"... What a primadonna. Anyway, it switches between the set of impulses depending on an input signal level. So you got the the soft sampled impulses when signal is soft and the other way around.
AA uses what is called the Vectorial Volterra Kernels Technology which implements an advanced dynamic convolution not a static one (or the "classic" dynamic convolution), like classic convo verbs do. Usual convo includes one big picture of the reverb you wish to recreate (the impulse response), AA uses multiples small samples to do its recreation, the sampling is split in small kernels and then it applies the same group of kernels to a group of samples and morph them, for the rendering part, in real time (ergo the big CPU hit). Basicely VVKT allows to record real analogue audio, more precisely where classic static convolution samples one picture, it samples streams of different intels with great details (volume, freq distribution, harmonic distortion). The samples are realized throught n.a.t 4, their sampling tech, you end up with a lot of small samples that are next used to shape the sound you process through it, giving it the qualities of the harware that was sampled with great precision. More detailled and with a vaster domain of application than traditional static convolution reverbs, in short.
how exactly? spectral morphing? so I can't use Volterra series in my own development w/o recording (sampling) real hw? Kernel is the same thing as grain in granular synthsis or some kind of windowed snippet of audio? So this is patented technology or I can figure out how to implement it by myself? I thought that there are just a lot of samples for accuracy
You have detailled questions, I wouldn't want to answer with incorrect informations, besides the basics which I've tried my best to explain, as I've understood them. I use the tech and I'm curious, but I can't detail things that deep, maybe you can find more in depth infos on AA website. For the sampling yourself part, it used to be possible with N.A.T 3 in relation with Nebula 3, but with N4 and N.A.T.4, the way I understand it it's only available to people who went through a formation of some kind with AA.
Sorry, yeah. It was the tool used for sampling for Nebula 3. Still is actually, at least I'm guessing. There are still libraries coming out for N3, and N3 is still available to buy from AA. N.A.T.4 is the latest generation of the tool, I guess we could say it like that. It came with N4. Here is how they explain N.A.T on AA website : "N.A.T. is the FX Sampler. It is able to sample the great part of audio signal processors like: compressors, equalizers, filters, preamplificators, guitar and bass amplificators, tapes, reverbs, microphones, Time variant effects and HRTF (head related transfer function) N.A.T. is based on VVKT too. It permits to sample the original linear impulse responses (traditional convolution), the original not linear impulse responses (original harmonic distorsions), the dynamic behaviour (for example compressors) and the time variant behaviour (for example choruses, flangers and generic time variant processors)".
