Also I’m happy to move to a different thread to continue discussions on how to do detailed analog modelling, so that Vult doesn’t have to worry about spamming the CF100 thread:
3 Likes
Response is not spamming, this is a fun little community with a lot of back and forth and threads that meander. No harm done, your nice work and vigorous passion are accepted arms open. It’s just unnecessary to get prickly, almost everyone is here for the right reasons, love of music made with fun software. Just enjoy the ride is my advice and meet some fun fellow designers and synth enthusiasts.
7 Likes
That is a shame that Andrew felt the clap took too much cpu. Would it be possible to add the fuller model as a right click “high quality” option?
1 Like
I appreciate that you want to go as close as possible to the original circuit. At the model level, you know that it is easy to capture a model of a circuit in a simulator like SPICE (or System Modeler) optimize some parameters and get very very close to measurements. Filters are very small electric circuits. Using the correct techniques it is possible to efficiently and precisely simulate those systems. All the science to achieve that is already in the textbooks.
There are many music software companies doing accurate emulations, like Softube, Arturia, NI, etc.There are dozens of Moog filter emulations that claim to match perfectly the sound. It’s a nice exercise to create an accurate Moog filter model. But for me there’s not much motivation on releasing it so there are dozens + 1 emulations. Why shouldn’t I be free to make modifications and create my own version? As long as I don’t claim that my version is a “Moog original transistor ladder” filter I think is fine.
We have different objectives: you release models of filters that match existing filters, I release whatever bent version pleases my ears.
In my real work we have do deal with creating accurate models for clients, because if they are not accurate enough there’s the risk of the thing making BOOM. But in the context of audio we can live just fine with my filters as they are.
There you have a complex and precise model that is not a Moog neither a Roland ladder filter. To give you some details, that’s my experiment using a BDF solver on a ladder of four non-linear single pole filters. Because I used a BDF solver the CPU consumption is variable. The point you picked with high resonance is where it consumes more CPU because the solver has to perform more iterations in order to return a precise solution and make the resonance sound nice.
My other ladder filter model is Nitrous. In that one I used simpler solver (trapezoidal integration if I remember correctly) therefore it is much lighter on the CPU side.
This is Lateralus with low resonance compared to Nitrous
this one is with high resonance
The Freak hardware module uses the same Lateralus model but without the BDF solver.
@andy-cytomic If your issue is that people thinks that my models are accurate emulations of existing filters, I reiterate it, they are not. I manipulate them at the circuit, differential equation and code level. It is not my intention to emulate existing filters. But please do not put into question my ability to make accurate models and simulations. I might be a sloppy coder, but I’m a good engineer.
9 Likes
Also I want to make it clear, Vult does great work. I appreciate his modules - also any chance of getting the Steiner Parker to go above 5.6khz please? They are generally stable and are useful, which is much more preferable to something like the current E440 model which tends to blow up pretty easily when in regular use.
3 Likes
I’m using an iterative solver as well, and picked the same “hard” case, and I’m coming in a decent chunk lower than the cpu you’re using, perhaps we can chat and exchange tips on how to improve efficiency? BDF2 or TRBFD2 or Gear or Trap all end up being pretty much the same in terms of actual cpu usage all being implicit methods, it’s not until you start getting into solving more than one implicit step at the same time in the matrix (edit: that the cpu increases significantly), as this more than doubles the entires.
Thanks for clearing that up, that is one of the main issues. The other for me is your claim of being able to do accurate models but then not delivering any for me to use, which is purely selfish of me. So if you are able to pull it off I would absolutely love you to do an accurate model of some good sounding analog circuit, make a VCV Rack module and a video showing the match, I’ll buy it! (I’ve already bought your other stuff too). Keep the warbles, keep the complexity and unideal behaviour, have a lower cpu mode if you must that strips out all that good stuff, but I’m fine with a high detail right click option!
3 Likes
I think you’ve pegged me wrong. I do my own designs and alterations too, I change things how I prefer them, especially in The Drop where I’ve got multiple different filters I want people to be able to easily switch between and compare the tone of the filters so I kinda cheat on the resonance curves etc so they match each other better, that sort of thing.
For example in the MS-20 mk2 I really liked how the circuit sounded not just when the OTAs were growling, but also when the OTA output buffers were clipping, but it took a lot of drive to get that to happen in the original MS-20 mk2 circuit, so I lowered the voltage rails on the buffers so they would clip earlier. But I made sure to match the regular circuit fist and post a video eg this from 11 years ago:
1 Like
For christmas, I’ll wish for peace and goodwill to breakout between all the talented developers. You are all very much appreciated, and it is a work very much driven by individual passions. The world is better off for it, even though sometimes it turns into heated tempers flaring.
I’ve had to spend much of my time at work this past month mediating between talented people fighting.
I was about to respond to Squinky, that I think the number of people with the talent and the interest to work on these kinds of projects is really still very small in terms of numbers, maybe enough for a very intimate conference.
It has always seemed to me that nobody can piss one off more easily than family, and I think the number of people who are passionate about developing new music instruments and new musical tools for people to use in our lifetime is so small as to essentially be a kind of family.
No wonder it’s so easy to fight over such little things! But I also think the relief at being understood and accepted is similarly great. It wouldn’t matter at all from a stranger, but from someone who actually understands what you care about and do, a little acceptance and appreciation can mean a great deal.
14 Likes
This thread rules. I love filters.
2 Likes