@Morphoice The basics are they are using ML models to analyze and replicate the behavior of analog equipment. They run various test signals through them and use the results to train a model that can predict how the device will respond to any input signal.
The Kemper Profiler uses a different approach. They capture the full response of an amp or effect at different settings. It creates a snapshot of the device and builds a profile. This is closer to convolution than ML.
The real problem you're going to run into with time-based dynamics (like a compressor with slow release) is the model will start to emphasize the non-linear behavior. The result of these ML compressors ends up sounding like saturators rather than compressors.
I'm on the same path as @griffinboy where I'm focusing on making models of specific components and inserting them between other analog models.
If you want to start tinkering with Neural models in HISE, here's a thread of my starting process:
https://forum.hise.audio/topic/8701/simple-ml-neural-network/108?_=1748359797127
@griffinboy also posted a thread about some details to the training scripts.
I haven't had time to fully dig in for the last few months, but plan to soon.

@d-healey I tried that recently just to reflect the bypass state of the DAW to the bypass button of the VST, but it wasn't reliable in the end... Maybe I've done it wrong that been said...

@d-healey

Or c++ / snex 😁

But two samplers is the easy way I'm sure.

@Morphoice scroll down in the export dialog, there's the juice.

@griffinboy I'm not at that point yet tbh. Most of what I'm working on right now is audio analysis in realtime and sample playback. I do really want to do a distortion/saturation effect of my own at some point though!

@Morphoice hmm, weird, anyways, I've added a safe check as well as the option to edit the flags set by the node_properties.json file in the DLL exporter. This should fix most issues related to this subject.

@orange yes some nodes seem to only work on windows and some only on macOS. The source code of my project can be compiled on windows and MacOS the same. Make sure to clear the binaries folder of the dsp networks. Do not delete any other folders.

When compiling make sure you first set all networks to not compile - then compile the dsp networks - open Hise and set your networks with Airwindows nodes to compile.

@Morphoice

Well maybe. But it'll be so cpu hungry. And fiddly because the routing of a real reverb is a complex matrix with taps coming in and out of all kinds of places

@Christoph-Hart sent

@JulesV Yes this.

Before I used a serial key system I would provide the reseller with coupon codes that provided 100% discount that they sell to their customers, and the customer redeems it on my site.

@Morphoice happens here too, no idea why

@Morphoice The solution I used was to keep track of all note on events, then if you ever receive two note ons in a row on the same note you know you've received some illegal MIDI.

It's pretty easy if on noteOn you check and store the event ID to an array (or MIDIlist I suppose) like

onNoteOn { if (events[Message.getNoteNumber()]){ Console.print(You've just received some illegal MIDI!); } else { events[Message.getNoteNumber()] = Message.getEventId(); } } onRelease { events[Message.getNoteNumber()] = undefined; }

How you handle the lawbreakers is up to you, but at a minimum it's best to stop everything on that note, as this situation breaks the assumption that every note on will have a corresponding note off.

In my case I'm using artificial events, so stopping them all is easy. I don't actually know how you'd go about stopping the normal events.

Give Synth.allNotesOff() a try, I know that doesn't affect notes with a timestamp in the future but I assume it works on stuck regular events.

It would be nice if we got this built in similar to setfixnoteonafternoteoff

@sletz Well, I just discovered that https://github.com/grame-cncm/faust/tree/master-dev-ocpp-od-fir-2-FIR. branch is still quite buggy, so no need to rush 😧

@Morphoice

You can map it to a function of best fit using python / matlab.
Ask chat gpt how to do this from a table of data.

Alternatively you could create a lookup table of accurate values.

I didn't realize you were going for analog tracking, that's a different case to digital, yes.
The tracking lookup only happens once per note, so you don't have to worry about efficiency here. Envelopes by comparison need optimization as they are applied per sample.

@aaronventure my bad, it was the detune thing from the example, triggering two notes at once

@aaronventure yes thats actually the one I was referring to it should be smooth already.

@aaronventure that certainly looks cool but would be inconsistent with the UI of the other plugins I made so far. Also my resources are very limited and I'd rather spend them on making a good sounding plugin right now and getting all the copy protection and distribution sorted before indulging in fancy lens flares ;)))))

@d-healey it's just intended as a nudge to upgrade, not a sort of copy protection. I guess those not able or willing to license a plugin AND able to crack/bypass wont be kept by any sort of method anyways, will they?

@aaronventure also held keys just stop when the sustain pedal is released, even though the key is still pressed. this happens with faust envelopes