Neural Quad Cortex

  • Nice video and it seems exciting, but let's wait and see what really transpires. Personally I think Kemper would be looking for a game changer to justify a version II hardware.

    I agree, I think Kemper nailed it out of the gate. I have tried everything out there except the AxeFX and I have not played anything else that ticks all the boxes like the Kemper in tone, variety and feel. That being said, if these guys actually pull this off and get it to market, it may be the first one that is able to go toe to toe with the Kemper. The profiling segment in that video sounded VERY similar to the Kemper profiling. We will see but I dont think it will be enough to make me want to sell my Kemper.

  • Profiling looks familiar because they most likely are using similar technic - fitting generic model to data. And model needs to "see" a veriety of data the learn. However I suspect that Kemper does it in a very smart way on slower hardware (most likely makes many well educated guesses in terms where it is _not_ worth to vary parameters while fitting) and neural probably employes machine learning fully taking advantage of modern, fast, multicore processors. My guess is they use transfer learning - they may be starting from pretrained neural network modelling generic amp and during profiling just train something small on top of it. In simpler words I suspect Kemper knows what it is doing under the hood, and Neural treats amp as a black box, automatically trained. Two different solutions but essentially same idea. I wonder if there will be any legal battle out of it because this method is patented by Kemper.


    anyway those are just my theories - may be off by miles :-)

  • I'm assuming every time they were A/B switching, both the Cortex and the reference amps were being monitored through studio monitors and not through a guitar cab?


    So proof of emulating a mic'ed cab, recorded sound, no emulation of "amp in the room" sound?

  • One of the things I do not appreciate when it comes to hardware, is when companies start promoting CPU-speed or all kinds of other specifications that in the end do not matter for the end user. It's all about the experience. Right now they offer a device with a 2Ghz Quad-Core cpu, but what does that actually say about the experience? It only worries me that the hardware will be replaced with a 3Ghz Octa-Core within two years, making the current Cortex 'not good enough' by then.



  • When it comes to traditional, "component" modelling, the CPU speed is a huge factor in determining latency. The more-detailed the modelling, the greater the need for a fast processor in order to render the modelled output within an acceptable time window.


    This is why it's a "thing" with the Fractal products (detailed component modelling), and it appears to apply to this unit as well. I remember form my Line 6 days, having owned everything up to and including the HD500, that as the modelling became more-detailed, the processors within the units were necessarily-quicker.


    As for how it affects the user experience, I agree that quoting CPU speeds means nothing 'cause it's the combination of the algorithms' complexity / CPU demands and the CPU itself that determines playability. It'd be more-truthful to say, "The algorithm is CPU-hungry and therefore demands a speedy processor, something we've managed to accommodate in the form of an x.xxGHz, xCore CPU, giving the player x.x ms latency".

  • Watched the capture video. Hard (for me) to say much with a high level of certainty considering the quick switching, playing variations, ect, ect...


    .. But I'm a little skeptical about how close this is to the source tones. Can be wrong, but I think I'm hearing something that may translate to a difference in so-called "feel" for quite a few. I'll need a unit here to test properly. May do a cortex vs kemper vs amp video at some point.


    I'll likely sell the kemper if I feel cortex is more accurate (even for non multiple distorting stage tones) in the low mids, some bass frequencies, without exhibiting something that may potentially be more problematic for me.


    Not sure what conclusions I'll walk away with though.

  • The way I see it with patents is what happened with Overloud and their profiling tech. They may have got around it by not giving the customer the means to profile and selling them profile packs instead. I wonder if Neural have something similar? I envision issues if they pack the tech into the Cortex unit... could be wrong though.

  • The way I see it with patents is what happened with Overloud and their profiling tech. They may have got around it by not giving the customer the means to profile and selling them profile packs instead. I wonder if Neural have something similar? I envision issues if they pack the tech into the Cortex unit... could be wrong though.

    Users will certainly be able to use the capture function themselves. About copyright, their approach is different enough, it seems to me.

  • Patents are only applicable when they are enforced. Neural DSP is another black box allegedly using machine learning but to what end is anyones guess, if it does what the Kemper does then it falls within the patent, if instead for instance it used machine hallucination to generate the resulting amp effect on the source signal then that might not fall within the patent. It's a different "black box".


    I guess we will have to see.

    I thought the video sounded great. If it can do distant mic'd sounds i.e. include the room verb/sound, or handle both power and pre-amp distortion together for accurate results then it might be serious competition even without the nice physical controls of the Kemper. The compact form factor would be advantageous for the road, but I'm not sure how fun it would be to program patches on the unit with just those touch foot switches or even with an app for your computer or phone - physical controls make a big difference when dialing things in.

  • If they find out a way to do the same the kemper does without the same technical basis applied, they can apply for their own development as a patent, black boxes are not patent friendly applications. The real question here is to what extent the quad cortex can profile the amps. To tell the truth I am not impressed by any of the sounds presented so far, being a model or a capture.


    So if the sound from the captures are good, nice, but how deep will they go into this? we dont know yet. Maybe we wont be able to separate cabinet and mic part from the equation, maybe editing will be pretty limiting, maybe dynamics are not as good as the real amp. We wont know until reviews arrive.


    Same black box or not, I wont believe until I hear. Form factor is fun, I would take it just for that if the quality is at least comparable to the big 3. (Kemper, Helix, Axe)

  • Mine is due to arrive in November.... but Neural have been shocking in the communications.


    I'm in the UK, and have been charged in Euro's for the deposit.... but the balance is payable in euros or dollars. How does that work? Can I decide which currency to pay in?

    Plus, it seems like we get it the same time as retailers..... so why would I purchase direct? What benefit have I gained from pre-ordering? We were supposed to get it a month or so early.

    The Facebook groups are not especially happy places right now!