Yamaha CS-80 VST

[mi-os]

Re the guitar/harp sound (whatever it is) from around 14:00 in the video. Did anyone manage to nail that sound on The Eighty? They sound so different. There is something off with mod/envs, idk. There always seems to be something wrong with envs especially.

The whole filtering section of the CS-80 is tricky because there are so many controls and a slightly unorthodox envelope.

As well as cut off and resonance for the HP and LP you have brilliance, keyboard control brilliance and initial brilliance.

Then the envelope has initial level, attack level, attack speed, decay and release.

The Eighty version of the sound in that video is way too dull and you'd have to play around with the initial level, attack level and attack speed to get that kind of ploink sound. To get more bass the usual trick is to set the HP filter very low and push the resonance.

I agree. There is a very different dynamic in filter modulation/envelope interaction on the CS-80. The Eighty misses that complexity and could be almost any VA. I woudn't be surprised if you could achieve the same sound with PolyM or PolyKB.

[zerocrossing]

Model 77 is quite soft, which is not the sound I associate with the CS-80. Great sounding synth, it just doesn't scratch this particular itch for me.

Is it not giving you the “Simply Having A Wonderful Christmas Time” sounds you’re after?

Like most of us, I’ve never touched or even heard a naked CS-80 IRL. Who knows what processing produced the famous sounds? All we can guarantee is that they were manipulated in some way on their way to the final product. I personally hate Vangelis and Toto.

I tend to mostly ignore claims of authenticity in any emulation. At best, it’s a “nice to have.” There’s an Arturia video with Steve Porcaro where he seems to basically nail his Toto sounds with it. I actually like Arturia’s emulation a lot, but Model 77 is still my favorite. All the modern CS emulators are pretty good, IMO. Each has their strengths and weaknesses. Best to own them all, so when Roy Batty comes a-knocking, you’ll be prepared for it.

[mi-os]

There a many youtube videos that showcase the CS-80. You can definitely get an idea of what it sounds like and what its special characteristics are.

[Uncle E]

Like most of us, I’ve never touched or even heard a naked CS-80 IRL.

Closest I've gotten is Deckard's Dream.

I personally hate Vangelis and Toto.

Now that you mention it, Model 77 and SoundPaint are great for Toto sounds.

[thedddkkk]

i like the gx-80, i dont like arturia. as with all their soft synths it all depends on the effect section. if you turn that off and the sound is gone. and it seems to me they use this effect section to cover up the imperfection of their soft synths. it seems under the hood all their soft synths are the same. they make a new gui, maybe tweak here and there and then they let everything depend on that effect section.

[EvilDragon]

The newer Softube Models (80 and 77), for example, are out of the question for me due to their constant CPU load. I refuse to accept there is no better way to code it more efficiently.

No, there isn't, if you want to properly model the analog device. Analog synths are running their oscillators all the damn time, they aren't stopping for anyone, amp envelopes are just gating the output in and out. So, to properly emulate the phase interactions between the oscillators (especially with modulation messing those phase interactions up), you need to run your code constantly. In that regard (and many others), Softube did an admirable job.

[EvilDragon]

I agree. There is a very different dynamic in filter modulation/envelope interaction on the CS-80. The Eighty misses that complexity and could be almost any VA. I woudn't be surprised if you could achieve the same sound with PolyM or PolyKB.

Faders and knobs have 10-20% tolerance on analog devices, and then there's device calibration. So the device that the guy in the video has most likely has parameter ranges different from the device that was used as a reference when creating any particular plugin. So you cannot ever depend on matching parameter positions 1:1, this will only work when comparing against the originally modelled specimen (and only if you're lucky, even).

Super important read: https://urs.silvrback.com/specimen

[Shabdahbriah]

Super important read: https://urs.silvrback.com/specimen

[mi-os]

The newer Softube Models (80 and 77), for example, are out of the question for me due to their constant CPU load. I refuse to accept there is no better way to code it more efficiently.

No, there isn't, if you want to properly model the analog device. Analog synths are running their oscillators all the damn time, they aren't stopping for anyone, amp envelopes are just gating the output in and out. So, to properly emulate the phase interactions between the oscillators (especially with modulation messing those phase interactions up), you need to run your code constantly. In that regard (and many others), Softube did an admirable job.

Computer code is not an analog circuit that needs to run all the time.

Simplified example for demonstration purposes: say you're running two sine waves at different frequencies. Stop running them, measure the time and use the delta to restart at the exact time they would have been if they had run all the whole time. Their phase relationship and everything else is excactly the same. In principle, that's certainly not the problem. Off the top of my (non DSP coder) head, nonlinearities and stuff like random numbers may make things a bit more complicated or even impossible. Another problem could be the additional latency, as the calculations of the entire model have to be restarted each time, at least for the number of voices played. There could be unwanted peaks, but not sure about that.

To claim that code cannot be creatively optimized is a bit deluded in my view. There may be solutions that take some shortcuts but do not introduce any audible drawbacks. What if one day someone achieves what was previously considered impossible? But of course i'd accept it if it simply wasn't possible for whatever reason.

[EvilDragon]

Simplified example for demonstration purposes: say you're running two sine waves at different frequencies. Stop running them, measure the time and use the delta to restart at the exact time they would have been if they had run all the whole time. Their phase relationship and everything else is excactly the same. In principle, that's certainly not the problem.

That method falls apart as soon as you add any sort of modulation to the oscillator.

[mi-os]

Simplified example for demonstration purposes: say you're running two sine waves at different frequencies. Stop running them, measure the time and use the delta to restart at the exact time they would have been if they had run all the whole time. Their phase relationship and everything else is excactly the same. In principle, that's certainly not the problem.

That method falls apart as soon as you add any sort of modulation to the oscillator.

That's why i said simplified. Modulations are just oscillators and processors, so the same principles would apply to them. All parts together form a larger equation.

[mi-os]

Regarding the best CS80 emulation.

I think when you take everything into account (sound, features and resource usage), Xils The Eighty is the best of the CS80 emulations currently available.

Model 77 is also good, but a CPU hog. (Edit: still think it's not very accurate)

[machinesworking]

The Xils Lab is completely useless to me, the GUI will probably go down in history books as to why following skeuomorphic design without regard to the user interface of a mouse and computer screen is a disaster.

[mi-os]

Let's say it doesn't win because of the GUI. If you enlarge it as much as possible on your screen, it's applicable. It could be better of course.

[twal]

Well in the video posted previously, I thought the Arturia sounded better than the Softube or Cherry Audio versions. The XILS sounded interesting and lively but perhaps a little saturated somehow. The Softube sounded plasticky but that might be a quirk of the what the guy was doing. I don't buy into the Arturia-isn't-good meme, to be honest.

Orbital and Underworld specifically use the Arp 2600 V. Perhaps the Korg sounds better a la Bones, which I never have heard; and I love the Arturia one's sound. It goes to show that Arturia gets too much hate for sure, when they do hit some nice, sweet spots.