KVR Audio

bronswerk · Post by **bronswerk** » Fri Jul 21, 2006 6:38 pm

I'm not sure how people look at Zyklus Improvisor or what it should be in their eyes, but the intention of Improvisor is:

1. Extensive midi master control functions like zoning, switching particular modules (arpeggio etc) on and of without note hanging, transposing modules like arpeggiators and pre-recorded sequences, see below.

2.Arpeggiators. These arpeggiators would be very powerful, can run on different clockspeeds and an arpeggiator's output can modulate another arppeggiator's input.

3. Recording notes in realtime for a certain amount of bars and instantly put that in a loop, ready for transposing etc.

4. Some sort of scale/note correction. When all these modules are run, there could be a problem of unwanted notes when transposing. See Bill's pdf file.

A standalone program sending and receiving midi only is in my point of view to most reasonable way to go at first. There are a lot of programming issues to be solved. If the program grows then other features can be added, but to make the above points work is a hell of a task already.

omissis · Post by **omissis** » Fri Jul 21, 2006 6:44 pm

Well said; the hardest part will be the real-time transposition logic which is derived of course by Direct, but I'm also thrilled by the possibilities of interactions between every module , relating to the master clock timing and its fractions, I can't see when the fun ends with such a creative tool!

bronswerk · Post by **bronswerk** » Sat Jul 22, 2006 9:00 am

lung wrote: Anyways, for now the solution is to create your own meta-instrument in a modular env., until maybe the OS movement or other inspired developer takes Bill's ideas and runs with them. MusicWonk is a very cool program btw that should be checked-out. I'm emailing the developer now to see if there is any possibility of it becoming a VST FX plugin, as well as having the ability to host VST's internally eXT style. I'll point him to this thread as well.

[Going a little bit OT.]
Funny you mention it, because that's my "plan B" I'm working on. I've created in Artwonk (same as MusicWonk qua midi, but with more graphic possibilities which I don't use btw.) a program ("patch" in Artwonk language) with 8 stepsequencers. Each stepsequencer consists of 8 steps, but these steps are not fixed in their position. They can be shifted left or right in their position, driven by a "shiftgenerator", which is a stepsequencer by itself. For each step you can set relative transpose, velocity, midichannel and mute. The ability to set a different midichannel per step is very cool. You can set start and end step, direction (left, right, alternate) for each stepsequence.

(edit: shift can be applied independently to relative transpose, velocity and mute. Not for midichannels because that's were I get hanging notes. But you have fixed midichannel per step, which is interesting too)

There are also 8 clocks. A clock is also a stepsequence. You can set duration, gatetime and skip per step. Multiple clocks can be combined into a "clockdriver" in a matrix giving more rhythmic variations for driving a stepsequence. And a clockdriver can be shifted too, so for eg a clockdriver uses clock 1 and 2 for four bars and the next four bars it uses clocks 2 and 3, and then going back.

Transposing on the fly happens by keypress on the keyboard. But you can enter a chord too, see below.

Then there's the biggest hurdle. When all these stepsequences play you can get a cacaphony of sound if you are not carefully program these stepsequences, in practice you are stuck with primes, octaves and quints. This gets boring eventually. So you must program the stepsequences more freely, but when shifting (see above) you get the danger of unwanted "clusternotes" that's for eg playing a C, D and E at the same time.

For this problem I use two mechanisms:

step 1. a (rudimentary)chorddetection system, for forcing notes into a particular scale eg major, minor, 7th, sus2, sus4 etc.

step 2. When scaling is applied there's still the danger of clusternotes. So I use "dynamic filtering" of notes. Per step the notes are "counted" meaning how many C's, D's whatever are going to be played. If notes are too close to each other, the note with the highest count is prefered, the unwanted note(s) is or muted or forced to rootkey. This is still in progress however.

Already I had some impressing results with this system. It's a perfect idea generator.

omissis · Post by **omissis** » Sat Jul 22, 2006 2:29 pm

bronswerk, one statement and one question.

First, I reckon that KVR is a fantastic place to meet the users and a large bunch of independent programmers who basically build stuff and compose stuff just for fun; sadly I'm realising day by day that this may be not the right place to meet motivated programmers who want to challenge themselves with professional stuff like the Improvisor; after all I see some big flaming threads about copy protections but fewer interest in other things, so I'll better suggest Bill to look over different places, and be sure I'll do my part to help him.

The question: I understand that your projects are underway and I'm sure you can have a personal approach to the problem of live composing but, having read your previous msg I'm wondering why don't you address your skills to the Improvisor tool.Basically you would get all the constant collaboration and further documentation from Bill so the designing and coding tasks seem to be not so terrible; moreover it could be largely appreciated if some of the very few programmers oriented on live composing who make up an active part into this board would get together in order to have Improvisor released for the public; maybe it's a dream, maybe not....

bronswerk · Post by **bronswerk** » Sat Jul 22, 2006 3:42 pm

Max, if I was an experienced programmer I would already definitely help Bill on the programming side of Improvisor. I do have some programming skills left over from my early Delphi years, but I'm an amateur/hobbyist and Improvisor is a daunty task even for experienced programmers. I feel I'm inadequate in this respect. Bill however is an electronic engineer with a lot more experience in these situations. I would go flat on my face so to speak. I'm simply not in the same league.

The Artwonk/MusicWonk patch I created has very few real programming issues; the language it uses is very, very poor compared to the more professional ones like C++ or Delphi. I really miss that. The main principle of Artwonk is you build your own patch from ready-made building blocks. Each block has it's own input(s) and output(s), where each block is doing its thing and pass their data to the next one. It's more like Lego. The patch is purely for personal use; it helps me to understand much more clearly the possibilities and problems when 8 sequences are playing together.

omissis · Post by **omissis** » Sat Jul 22, 2006 10:17 pm

Anyway more skilled than me that ain't got a single knowledge in programming! Why not proposing yourself? After all it's not DSP pure, it's just to create some logic functions which get a special kind of routing and all that!

omissis · Post by **omissis** » Mon Jul 24, 2006 9:34 pm

No interest? Too a hard project for your head, guys? Holger? Jørgen? KVR best brains, where are you?

WMVyla · Post by **WMVyla** » Tue Jul 25, 2006 10:20 pm

I want to thank everyone for their support and suggestions. It is really strange that no matter how many people think an idea has some merit that the major manufacturers don't want to know. I really don't understand this at all considering I never had any intention of making any money from the project. But I guarantee that if I did get a product out there, we would see many parts of it being copied... It is about time the industry woke up anyway. The way things are the the moment...well, how many channel strips or mic preamps does a man need? I counted 27 and 38 respectively on a quick surf of the net. And just don't get me started on the number of compressors. Of course, these things are useful, but what we need is something new and fresh for the creative side and not more endless add-ons for the recording side. Something that would appeal to both beginners and accomplished musicians. Something that can also be fun. Something that doesn't need a four metre wide monitor just to show all the windows on the screen. Oh well, that is my rant for the moment.

In the last few days, I have been contacted by a major keyboard manufacturer who are interested in the idea of me designing a master keyboard somewhere along the lines of the Keyboard Controller spec (also on www.vyla.co.uk). This would have all the stuff required by a pro musician for use during live performance and thus would not have the usual mountain of soft synth controls. It would have a few, and the configuration of these would change depending on the song and what synths are in use during that song. It would have all the usual zone on/off stuff on separate buttons, names of sounds as text rather than numbers, etc. plus some organisation stuff that would make it an alternative to NI's Kore. It would also contain virtually all of the Improvisor functionality as well. The best bit is that the price would likely be under 500 GBP for the full thing with an 88 note hammer action keyboard. Other variants would be cheaper. I can't say the name of the company, but their current 88 note keyboard has a metal case, a lovely action and feels like you could drive a truck over it without breaking anything.

This really is the last ditch attempt to get something going and I am quite excited about it. Obviously, I cannot go into any details about the spec that is being discussed but suffice to say that the only things from both these specs that won't appear in the design are things that simply cannot translate to hardware. If these discussions fail for any reason, I will try to find time to put a working framework for Improvisor out as an opensource project so that new modules can be added by other users. I don't quite know how I could do this yet because the end result should look like something like Improvisor and not like SynthEdit if you see what I am getting at.

However, for the moment there is a bit of hope for this project on the horizon so I will see where this leads. I hope that the potential move to hardware doesn't annoy too many of you.

Many, many thanks again for all your support.

spuddle · Post by **spuddle** » Tue Jul 25, 2006 11:06 pm

Wooo, new CME keyboard with ultimate control on the way I take it!

Can't wait.

omissis · Post by **omissis** » Wed Jul 26, 2006 7:50 am

Bill , I could nothing but be happy for you , of course you know this is a preliminary stage so nothing is really clear here yet, anyway I'm happy to hear there is somebody who's really interested.
I mean, the improvisor is a wonderful machine still a bit ahead of its times! Not everybody can realize what a spontaneous way of composition made in the same moment of recording is, so it's wonderful that this project will spread this new way of composing through the users, even the newcomers.
Again, if you need any help which is not code-tied I'm here!

bronswerk · Post by **bronswerk** » Wed Jul 26, 2006 1:56 pm

Congratulations, Bill. Finally someone is definitely showing more interest in this project. Man, a hardware keyboard with a lot of the Keyboard Controller/Improvisor features is in mine opinion a killer keyboard. I would buy it instantly. Of course we must wait to see how things work out, but for now everything is much more positive than let's say a week ago.

dlmorley · Post by **dlmorley** » Thu Jul 27, 2006 12:16 pm

As a Zyklus MPS user, this is good news Bill.
Hopefully the keyboard can get off the ground. I'd be on the list.
David

lung · Post by **lung** » Thu Jul 27, 2006 4:19 pm

Excellent news. Congrats Bill. Wish you the best with this new development. I'll be right in line after David (love your music/sounds btw, welcome to KVR). FR: possible USB/VST editor so things like automation, dynamic loading, save presets in songs.. can be achieved. mLan for less MIDI cables, yet multiple MIDI streams to the computer/VSTs. Very excited!

omissis · Post by **omissis** » Thu Jul 27, 2006 6:04 pm

Bill
I really hope the project will dock somewhere also in software form, just to be also free from an hardware like it happened for K.A.R.M.A. , although not linked to a specific keyboard

maki · Post by **maki** » Sun Jul 30, 2006 6:25 pm

Hello!
I am not a programmer but one who enjoys music and everything that has happened since Vst standard was born.
I read thread about realtime midi manipulation and Zyklus and Direct. There is one thing running through my head: what is more important-to emulate Improvisor or to have all those options available to end-users?
I think that todays modular enviroments like EnergyXt does some, if not many of abovementioned things. And if we think of all those free and commercial midi plugins available today, than I must admitt that possibilities are tremendous.
Especially in modular enviroment. Vangelis did not have that much freedom, did he?
However there is one thing still not available at all, not even in a rudimentary form.I am talking about your 4. statement. We do not have any plugins that does scale correction or basic auto-accompaniment. We do have
external programs that emulate traditional one-man-band keyboards with styles. But there is no such a small, simple modular plugin that could transform midinotes in realtime according to predefined rules or even simplier,just forcing to notes user inputs while playing.
There are many midi loopers out there, even phrazal samplers Vangelis could hardly find 15 years ago. There are many converters like midimunger and plenty stuff from soma, nic-fit, tobybear, NCD and subminimal to play with. And most of them are donationware! There are also acceptable remote controll
posibillities to controll those midi flows with hardware controllers.
To make a long story short, somebody could emulate that Zyklus 100% one day, but why not put focus now on few elements we need to make realtime performance more interesting.

I am suggesting some kind of cut-down version of autoaccomp midi plugin that processes midi-notes through an auto-accompaniment algorithm.
But it would not be just a "light" version of Band-in-a-box but a new tool because it could be used differently than traditional auto-accompaniment.
That modular autoaccomp plugin should have options to choose at least between bass and "other instruments" algorithm because those track are processed different in an auto accompaniment system.
There would be no need for styles at all and no need for any presets at all if host could save everything as a part of a modular project.
Plugin could trigger autoaccompaniment with midi messages from hardware controller buttons like we know on hardware systems-for fill-ins, variations,intros or endings.
That plugin should not play styles at all... It should be a new composing tool, simple but functional good enough for modular purposes. If user needs more tracks, user could load more plugins with different midiflows and perform many other things not previously possible with traditional auto-
accompaniment programs.
No nedd for such a tool to have any user unterface at all. At least in first version.
For users who like to play with modular components and love experiment, this should be also a new creative, magnifficant live-performance tool, never seen before in that form despite fact that those auto-accompaniment princips are now 25 years old.

Regards Maki

developers.. live MIDI performance VST (Zyklus/Direct)