Do I “need” analog synths?

[Jace-BeOS]

right, but 32 is nothing. I know vst's that need several hundreds of assignments.

what I don't understand is, what is a 14-bit midi knob supposed to be able to do or solve that an automation-wired knob can't, and since automation control works so universally well, why even consider midi at all?

How else are you going to connect that knob to the automation? Or do you mean an on-screen knob?

[Jace-BeOS]

IMHO i believe you need a combination of analogue for warmth, digital for cold and real recordings for uniqueness

The “sound of analog” is not a topic I wanted this thread to include. There are so many threads on so many forums that have that topic. Let’s please not go there. My interest is in the control and play experiences.

[Jace-BeOS]

You need it to convince your self that you don't need it.

This is a possibility I was sort of considering. Imagining using a thing is always different from the actual experience.

[acYm]

well, when you use a proprietary system, be it Kore or Automap or else, you bypass the midi protocol completely in the first place.
of all my templates, only one uses MIDI messages: Nerve, because I needed a way to assign notes to buttons. to accomplish that, an auxiliary Reaktor ensemble with midi out is used.
once you're in a DAW, the whole MIDI system becomes quite redundant... it can have its uses but only as a workaround, when a plugin doesn't have comprehensive enough automation support.
Kore's knobs aren't limited in any way by the MIDI spec (not in Kore mode anyway, it also has midi mode which is another story of course) and so it can access the full 16-bit worth of values that a vst2 parameter can have.

[D.K Envelope]

i like analog gear just for the workflow it feels good

[JCJR]

Aren't vst parms float rather than 16 bit? IOW lots bigger resolution than 16 bit? Or maybe ya'll are talking about some other vst data construction. Been awhile since I studied it and can't recall, too lazy to look it up.

[acYm]

that's what I remember from long ago... I'm not sure whether it's different between vst2 and vst3, or if there's a way around it... but 65535 values is a fair number, you could make a filter have 3 increments per hertz, that's not too bad. but that's rarely seen... more likely to have a vst where ever param just has 1000 steps on everything and such

[S0lo]

Most of these native systems like Automap and the like use MIDI messages internally. To check for that, one can use a MIDI monitor vst (or if you have monitoring features in your DAW). Even when the system is hardly native they tend to use Sysex which is governed also by MIDI. If it didn't then the DAW wouldn't be able to route the messages to the plugin. The plugin would need to get the messages by other means like TCP/IP, IPC or else.

VST internally uses float (32bit) or double (64bit) depending on the what the dev wants. Thats useful for internal modulation like by LFOs and such or DAW automation. Unfortunately it's not useful for external modulation via hardware (even when it goes through DAW automation). These are still dictated by the lowest denominator, in this case 7 bit MIDI or 14 bit MIDI or else.

[JCJR]

that's what I remember from long ago... I'm not sure whether it's different between vst2 and vst3, or if there's a way around it... but 65535 values is a fair number, you could make a filter have 3 increments per hertz, that's not too bad. but that's rarely seen... more likely to have a vst where ever param just has 1000 steps on everything and such

I used to program vst hosting among other things but retired a few years ago and memories may be faulty.

For a moment forget softsynths-- A great many FX vst plugins are not programmed to know anything about MIDI. So far as I recall the numeric parameters exposed to the host from the plugin are floats, and I seem to recall the typical value range of a parameter is 0.0 to 1.0. If you want to stretch the range to fit over maybe 0 to 20,000 Hz or whatever value the parm represents, it is simple scaling, but because a float has rather fine-grained resolution, the container is usually adequate to the task.

Consider a simple delay plugin which completely ignores MIDI messages-- It is somewhat host-dependent of what can be recorded or automated, but many hosts can create automation tracks where you can automate that dumb-ass simple delay time parameter, or some hosts can go into record mode and record automation tracks, recording your mouse gestures as you twiddle the delay knob on-screen during overdub.

Maybe some hosts would decide to record such data tracks internally into bytes or ints or whatever. Maybe some hosts would decide to fake the parm automation as MIDI controller tracks so they can be handled the same as MIDI tracks. Dunno. But so far as I recall the communication to and from the plugin are floats, so the simplest thing for a host to do would be to record and playback the same floats rather than trying to mangle them into something else.

I never paid attention to "knobby control interfaces" totally ignorant about them. If some of the knobby interfaces, you can program a knob on the controller to automate dumbass FX EQs, compressors, choruses, zillions of other plugins that don't give a hoot about MIDI and completely ignore MIDI messages, then somewhere along the line the knob gestures get turned into floats before they make it to the non-midi VST's. Maybe controller driver dependent or host-dependent or something else. There are many ways it could be done and still fit the VST spec.

Hard to know what data formats might be "in-between" maybe a lot of knobby controllers just send MIDI and the host or controller driver or whatever turns low-res MIDI into floats, in the cases of software/hardware combinations that can automate non-MIDI VST's. I'd guess there's maybe 10 or 100 non-MIDI VSTs for every MIDI VST.

Tis hard to say, folks been arrogantly trying to "improve" MIDI for decades, mostly just making incompatible ephemeral messes on-top of a crude but very viable framework. For instance back in MacOS 7 or earlier, late 1980's or early 1990's, Apple arrogantly megalomanicly released "MIDI Manager" which they thought was a vast improvement over MIDI. Basically dumbass stuff like keeping all the message types but using longs instead of bytes for data containers. And adding useless enhancements nobody used in the real world. So to program MIDI via MIDI Manager you had to reformat everything to their dumbass "new improved" structures on the way in and out.

The whole silly thang died on the vine in only a few years but because some software continued to use that layer, MIDI Manager had to be supported for a decade or more after apple dropped it like a hot potato. What a wonderful improvement to MIDI.

Thats just one example. Folks been "improving" MIDI ever since, making one heck of a confusing mess of it.

[acYm]

Most of these native systems like Automap and the like use MIDI messages internally. To check for that, one can use a MIDI monitor vst (or if you have monitoring features in your DAW). Even when the system is hardly native they tend to use Sysex which is governed also by MIDI. If it didn't then the DAW wouldn't be able to route the messages to the plugin. The plugin would need to get the messages by other means like TCP/IP, IPC or else.

VST internally uses float (32bit) or double (64bit) depending on the what the dev wants. Thats useful for internal modulation like by LFOs and such or DAW automation. Unfortunately it's not useful for external modulation via hardware (even when it goes through DAW automation). These are still dictated by the lowest denominator, in this case 7 bit MIDI or 14 bit MIDI or else.

obviously, internally there's a two-way communication system between the hardware and the host, so that it can work... but MIDI being so inherently limited, I just can't imagine why they would have used it. tying a parameter to a knob isn't something that absolutely requires midi to be involved. any other protocol would be more solid. and no message is being routed to any plugins... I'm controlling the automation just as directly as any host's timeline would. which would be why it feels as solid as it does...

I don't see any midi activity on Ableton's indicator when using Kore, Maschine or Automap, unlike all my other controllers which are based on midi control scripts, as you'd expect...

like, technically, Ableton doesn't even know that my Kore hardware is connected. Live is merely my multi-track recorder and motion sequence manager, whereas Kore acts as my actual daw... yet it's very much "just a vst"

[acYm]

epic novel

hmm. interesting. not sure why I'm under the impression automation values are limited to 16-bit then. maybe it was for the old vst1? or maybe I just read some nonsense back when nobody knew wtf they were talking about cause modern surfaces were a new phenomenon.

all I know is I've never, not once, not been able to access every value of a parameter if I wanted.

let's take mono/poly, where something weird can happen... it goes 0.00 to 10.00, that's the values you see in the daw. so you think it has a resolution of 1000. each of these values is seemingly assigned to a frequency...
1.15 = 30.337hz
1.16 = 30.513hz

and so on (it gets exponentially wider as you go up, which is fine), but if you hold shift and turn the knob with the mouse, and look at the pop-up that appears where you can see the frequency, you see that 1.16 can also be 30.690. that would be the vst's internal smoothing system at work, made so that automation can ramp smoothly... so of course when you turn an automation knob you get all these inter-values as well. so even if your knob rotation isn't exactly wide enough to physically see every value get stepped through... by making the same small movement, you'll never exactly land at the same spot twice, or, not too often at least.

[JCJR]

Here ya go--
viewtopic.php?t=359767

If a plug-in wants to present a text box or slider maybe it goes from 0 to 10 in 1000 steps or whatever, it does mapping of that range on top of float 0 to 1.0. You can only select 0.0, 0.01, 0.02, .... 9.98, 9.99, 10.00 using that plug-in control because that is how it is programmed but the host could send 1.02567 or any other fine grained float value. Because they are floats not every possible decimal number has a perfect float representation but there are LOTS more possible values that can be passed than the 1000 which the programmer had decided to let you enter via the control on that case.

Depending on the usage of the parameter, the programmer may decide to quantize "In the cracks" values to those he wants to allow. Or in other cases maybe he will accept any legal fine grained value and the only reason he limits the choices in his onscreen control are so that the GUI will look and work better. Maybe he doesn't want to expose big nasty text boxes where you can type-in and display 12 digit numbers.

Then there are "very limited parms". Maybe a sync switch where any value less than 0.5 means sync off and any value greater than or equal to 0.5 means sync on.

[Jace-BeOS]

I guess part of my query has been answered now: if I want high-resolution control over parameters by hand (not via LFOs or envelopes), there are only two options:

1. Abandoned software (more menu diving and mousing for setup).

2. Analog devices that aren’t digitally controlled, and cannot be controlled any other way.

I still need to decide if I want to spend my [rare] money to find out if I enjoy working with truly analog gear. :-/

[acYm]

in an ideal world, when ableton made Push, they would have made it a lot more like kore or automap in terms of universal vst control, so that it would all happen at the main daw level. having a sub-host seemed like the best solution back when every daw was equally ill-equipped in this department, but does make the whole process a bit more fiddly than usual, especially at first.
you can't expect to be quick at navigating from the controller and making chains so soon... but if you keep at it, next year it'll be seamless
I'd suggest loading up with various sequencers and using Kore2 standalone for a while... there's many more good vst seq's than there used to so it became great fun to run Kore daw-less, with looperator to handle the "project automation" while the main parts are tweaked on the fly

[vitocorleone123]

I've been working "in the box" for a long time. A couple months ago I purchased 2 synths: an Ultranova (mostly as a MIDI controller that fits my space) and a Roland SE-02 (mostly to help teach myself to get a better feel for subtractive synthesis). I love the SE-02 and just got the Ext Box expansion for it. However, I'm still struggling to this day to reconfigure my workflow to incorporate hardware - you can't just drag a new instance of the SE-02 to a 2nd and then 3rd track and keep going.

I do WANT more hardware synths, so that's something. Ideally, my next would be either the Peak or a Rev2 desktop. But unless and until I can actually finish some songs that incorporate the hardware, my money may be better spent on Omnisphere, instead. Only time will tell, but it's a toss-up for now.

I guess my advice, such as it is, is to try hardware, with the caveat that 1) it's probably not magical , and 2) it'll take work to make it work for you (probably), and 3) if you do love it, you'll be tempted to break your bank account to get more more more

Final tip: get something you can control and automate really well with MIDI if you're used to working ITB as it both helps bridge the gap and also if you drift away from hardware you can still treat it as a real virtual synth ( ) in the future if you don't sell it.