Nextmidi Strings Library? Midi tool? coming soon

[Guy Rowland]

This software is only useful if you have a fixed template of sounds to work with, isn't it? And I don't do that. I usually start with a few ideas, played with some sketch-patch or a piano, and then go looking for sounds which best express those ideas.

I suppose there's a case to be made for breaking the habit of a lifetime. Perhaps not working with full templates, but instead subsets for specific purposes. After all, conceptually there's not much difference betwen choosing, say, Sable and Magnificient Strings, if Magnificient Strings is effectively your custom-made sample set. And if you route the sample mics into the DAW, you'd have complete control of spatialisation etc as you need it on a case-by-case basis.

But otherwise - yes, this would be the opposite of a piecemeal approach to writing, and a quick way to stop a creative flow stone dead if it meant setting up something bespoke from scratch every time.

It would sit quite well for me in my disabled VE Pro template. I just load up Magnficient Strings at a touch of a button, as opposed to Sable.

[wst3]

I was thinking exactly the same thing Piet - well, I was actually wondering just how much work is required to configure for a specific template - would it be practical to create one config each for several "starter" templates or would this be the force that drives me towards a single template?

We will learn more, I'm sure...

[Luciano Storti]

Good ideas there, Guy. Get in touch with the team at some point, they specifically state they want to hear from (potential) users to improve the software, as long as it's feasible as a feature of course.

Piet, I would agree that in your case it would be a tougher "sell." It sounds to me like you would have to enable one Logic template with the environment setup required to make DIVISIMATE work, then save a "generic use" template in DIVISMATE itself to stay as modular as possible. Don't have enough information yet, but from what I've seen, it seems possible.

Question is, do you need it? I really see the strength of this in saving time/passes. Obviously, we can all call up AM/SFA strings and add them in the configuration listed above, then massage things as we see fit. This simply lets us do it for all of them at the same time. You might want more individual control over your parts...

[FriFlo]

IMO this is a great tool, but I already programmed stuff like that (polyphonic legato playing and routing) in Max MSP. And a tool like that is always much better, when you can exactly define the features for your personal workflow! Further, I plan to integrate this into my template control which is also written with Max MSP. I had kind of hoped, that this new tool will be kind of both things: A control surface for your template with labeled key switches, that knows automatically which instrument track is selected currently and provides you exactly with the controllers that are available for the given track. I simply don't get why on earth nobody thus far has managed to offer something like that in its glory! The only hopeful attempt I have seen is the Patchboard App by MR Miller, but the website has shut down, so, I suppose he returned to making it a product for the few composers that are willing to pay several thousand dollars plus maintenance, like Danny Elfman or Brian Taylor who are his customers ...
Anyway ... at some point I will have programmed my own version to the full extent and then I might try making that a commercial product. I had always thought, before I manage to make that happen, somebody else who is a real programmer and has more time for stuff like that will be quicker then me ... seems like I was wrong about that.

[FriFlo]

And some observations that may not be so obvious to people who have not mind-fiddled with the idea of polyphonic legato playing:
Every way of doing it has some disadvantages! The basic problem is always this: let's say you want to divide you playing on a Violin, Viola and a Cello. In order to know where to route the a note, the engine has to know where to place the voice - top, middle or bottom. In order to decide that it has to know the full chord with its three notes to decide where to put it. That means there needs to be a delay between your actual playing and the notes being played. But that is not alway how polyphony works! There might be the violin starting alone and the other instruments set in in a later bar. Or there might be one instrument playing the same note as another for just one moment, or voices do cross! Or ... or ...
If you keep thinking about a way to make all of these circumstances possible you quickly realize it simply is not possible for all circumstances of polyphony. You can find different ways of a script working for different circumstances, but in the end you will need to go into each individually anyway, unless you don't want to narrow your composing to what such a tool can do.
I still think it is very useful and inspirational to play on it, which is why I still work on my own version integrated into my template control. But I suppose many potential customers will only start seeing the limitations of such a tool after buying it.

[Guy Rowland]

Yes, very true FriFlo. I'm a bit worried about that aspect of it, divisi handling is always a bit tricksy. Audiobro seem to do it better than most in terms of "just seems to work most of the time", but that isn't 100%. I guess what's interesting about this approach is that it writes everything to a separate track, so you can go back and manually fiddle to get something right it auto-detected wrong.

I'm intrested in your controller idea... I don't 100% get it yet, but it sounds intriguing.

[Guy Rowland]

...and one more thought. On the issue of the Strings demo that we all fell in love with... what it seems to come down to really is layering Mural (or other nice ensemble library) with three instances of Audio Modelling solo strings in the right proportions. Is that it?

[Scoredog]

I would think you could do this with any solo violin patch. Those with Afflatus can achieve a similar result now too...but that is just a custom multi and this thing has the whole switching deal too.

[Guy Rowland]

I would think you could do this with any solo violin patch. Those with Afflatus can achieve a similar result now too...but that is just one patch and thing has the whole switching deal too.

Hmm, it definitely sounds different to the ol' solo violin trick. I'm guessing that the 3 solos all need humanising in some way - detuning especially.

[tack]

...and one more thought. On the issue of the Strings demo that we all fell in love with... what it seems to come down to really is layering Mural (or other nice ensemble library) with three instances of Audio Modelling solo strings in the right proportions. Is that it?

That was my reading of it. I'm surprised, but if it's just that, it's a great example of blending libraries being (much) greater than the sum of its parts.

[Piet De Ridder]

(...) Question is, do you need it? (...)

I don’t think I do. See, the first thing that comes to mind as a major problem, for me at least, is — and this adds to the concerns that FriFlo already expressed — that Divisimate seems to assume that all libraries are consistent and reliable, and that they all do well what they claim they can do, across both the instrumental range and the dynamic range. Thing is, I don’t know of a single such library.

For example, when I program spiccati or pizzicati or short brass or woodwind notes, I have to manually adjust dozens of individual notes to compensate for inconsistent attack times and such. Or sometimes there’s a single note or odd legato-transition that jumps out for one reason or another and that needs individual care. And that’s just two examples, but there are countless of situations — with all orchestral libraries — where you have to go in and give individual notes or transitions, or certain samples in a round-robin-series or whatever, some extra meticulous context-dependent attention. As far as I can tell, that’s not something you can leave to Divisimate, ingenious piece of software though it undoubtedly is.

Another thought: I often assemble performances of, say, a solo bassoon, horn or oboe or whatever by combining different libraries, picking notes and articulations from whichever library gives me the sound or performance that best fits whatever it is the part needs at any given moment. So, in bar 1 the note E4 may come from library A, but in bar 2 that same note may come from library B and a few secons later, it might be a different library (or modelled instrument) again that provides the sound for E4 … All these choices are, again, very context-specific and as such decisions that I can’t, and won’t, trust scripts or clever software with.

I may be completelty wrong, but it seems to me that if you go the Divisimate-way, you’re more or less resigned to accept a sort of generic, passe-partout performance from your libraries. I don’t question that it can be a great help in certain set-ups and with certain workflows, but I work completely differently.

I could actually do with a Frustrationmate instead of a Divisimate.

_

[Guy Rowland]

Your way of working is fascinating Piet... do you mix and match in that way for ensembles?

I see this as a sort of bespoke alternative to an ensemble library, or even bigger, a Symphobia-like Multi. Which, its true to say, is probably not classic Piet De Ridder territory.

[FriFlo]

I completely agree with Piet‘s reasons what this software does not help with. But I still think a divisi tool like this can be very helpful in spite of that - there is no contradiction. What about playing in a crude idea half decently and then later replace most parts with refinement? For the usual TV production where you have not much time on top of that (because they don’t pay well enough, that will have to do and that is a great tool for work. But if there is, nothing will stop you and the tool will have helped you finding the right Orchestration. I might add that my own version can not only divide a breath controller or other CC to all notes, it can also use poly pressure or mpe to sculpt the vibrato and/or dynamics of individual notes within a chord. It can combine that with a fader or expression pedal or Breath controller and give a curved for the CC resulting for the instruments individually.
My reservations towards the software are more in that regard: I think there is way more stuff that can be accomplished and for a system that I have to adjust my workflow to, I would require that amount of detail and completeness (if that is a word?).

[Ashermusic]

If either my clients or my conscience required me to go through what Piet does, I would want to end my life But Piet has an unsurpassed dedication clearly.and I totally respect that.

But for me: They are f’ing samples and instrument models! If I am forced to use them instead of real players, as opposed to adding them to real players, (and yes, Daryl, I am FORCED to or I can’t pay my bills,) then I want it to at least be moderately enjoyable composing with them so am not going to make myself crazy spending hour after hour in the quixotic attempt to achieve even close to parity with real players.

So, if this tool actually turns out to be something that achieves reasonably good results in a less laborious and soul sucking way, I am interested.

[mcalis]

I thought I'd share some of my thoughts after having played around with the software for a few hours last night. This isn't a full review by any stretch of the imagination and there's a good chance my impressions will change as I familiarize myself more with the software.

First off, the installer is quite small, installation is quick and easy and I can confirm it works as intended on Windows 7 64-bit (which isn't necessarily a given anymore these days). The GUI is minimal and to the point and with the aid of the video's released so far by nextmidi it wasn't difficult to figure out how to hook it up. The only confusion I still have is that I didn't yet manage to find a way to route a voice/note to a specific channel on one of the MIDI ports within the DivisiMate interface. I can change ports in the matrix view, but I do not seem to be able to route a voice to a specific channel on that port, which seems like a strange omission. Effectively this meant I had to set up CSS Vlns 1 on DivisiMate Port 1, CSS Vlns 2 on DivisiMate Port 2, etc... whereas I figured it ought to be possible to have CSS Vlns 1 on DivisiMate Port 1, Channel 1 & CSS Vlns 2 on DivisiMate Port 1, Channel 2 etcetera.

I'll have to inquire with nextmidi to see if the aforementioned is possible, because if it's not, that would be a rather severe limitation, especially if I want to use DivisiMate with stuff like Hollywood strings which has minimal keyswitching support and no articulation-switching-by-CC support either.

DivisiMate allows the user to set up a trigger for each of the MIDI ports/target instruments when a DivisiMate preset is loaded. Basically the use here is that you can send a CC or keyswitch in order to select the right articulation on the instrument. Bizarrely, the range of key notes that can be send only goes up to A0, meaning I couldn't send C0 + A#0 to CSS in order to select legato. Fortunately, CSS also support articulation selection through CC58 but it seemed a rather strange omission regardless (also, selecting notes and the velocity at which you want to send the note is done through very tall dropdowns which doesn't make for a great UX, I'd rather be able to type in C0. Then there's also some weirdness that some notes in the selection are written as sharps, and others as flats whereas I think MIDI convention is to write everything as sharps).

When I started to play it became very obvious that having a breath controller is pretty much a must in order to get the most out of this software. Trying to play a chord + melody with my left hand (or as Guy would put it: boxing glove) and riding faders with my right at the same time proved rather difficult. The software can't be blamed for this though, but I do feel that in order to really get value out of it, you need to be able to have both your hands on the keyboard (If anyone has good suggestions for bang-for-the-buck breathcontrollers, I'm all ears!).

I was kind of hoping there'd be a way to offset CC values within the software (in other words: being able to lower incoming CC1 for voice 2 and 3 by some fixed value and boost it for voice 4 and 5 while leaving voice 1 unmodified). Currently you simply get homogeneous CC data across all tracks you record to, bar some of the randomization/humanization you can specify in the settings (but these are global settings, not per MIDI port).

There is also the option to use a repeater on MIDI ports which would basically allow sending a rhythm to a MIDI port (I know there's a name for this stuff, but my morning brain is failing me). Haven't tried that one yet.

So where does it shine? Well, I can definitely see value in using the tool as a way to get ideas down quickly which, to me, is valuable. I am less sure if you could use it effectively in a full on production where you might switch between DivisiMate presets every other bar, but that's something I'll have to try. Still, the value of a tool that allows you to essentially construct your own multis from whatever libraries you have available is useful, though my feeling is I would mostly use it as a sketching tool to get an idea down quickly and expressively. "Sketching tool" might not be a moniker that sounds very appealing, but that's where I see the most potential currently. Of course you could also use it to write detailed inner voicing divisi parts... but I am not sure if you'd really be any faster than if you simply dialed in a few seperate lines. The strength is definitely in "tutti"-style playing.

Lastly, I see pretty great potential combining DivisiMate with the cinematic studio series, because those libraries are very well matched to each other and use the same keyswitches/CCs, etc. The other obvious choice is to combine this with modeled instruments, of course, but I don't have any of those in my arsenal currently.

Some of the concerns raised here by FriFlo and Piet are definitely applicable and on point. This software won't serve every workflow, but if you like your Symphobia multis... this might be an interesting tool. For people who are tentatively interested, I'd probably wait a little while until the software has matured more, is more feature-rich and until the early adopters have figured out the best ways to make use of it, especially because it does take a little while to set up. There's also a real possibility that users might start sharing their templates further down the line, which could be a real time-saver for people who decide to pick it up later.

[Guy Rowland]

Matthias, huge thanks for that excellent overview. No keyswitches below A0 seems quite absurd - all mine are below this, pretty much. Also I definitely know where you're coming from with the running-out-of-hands thing.

For me the bigger thing is simply that sound they got with the strings, which seems to be about mutliple modelled instances alongside samples. My initial interest is 90% that and 10% the app.

[mcalis]

yeah, I don't know why they did that. Here's a screenshot...maybe it's there but I am missing it.

[mcalis]

Matthias, huge thanks for that excellent overview. No keyswitches below A0 seems quite absurd - all mine are below this, pretty much. Also I definitely know where you're coming from with the running-out-of-hands thing.

For me the bigger thing is simply that sound they got with the strings, which seems to be about mutliple modelled instances alongside samples. My initial interest is 90% that and 10% the app.

I accidentally wrote that it only went down to A0 but I meant up to! Morning brain...

So yeah, most Keyswitchable arts should be covered by this, but not CSS/CSB.

You can save these trigger values as presets though, which is neat.

[tack]

An alternative to a breath controller could be an expression pedal. I use one and assign it to CC1. It's let's exhausting than a breath controller at least.

[FriFlo]

An alternative to a breath controller could be an expression pedal. I use one and assign it to CC1. It's let's exhausting than a breath controller at least.

If you find a breath controller exhausting, you should write more rests! ;-)

[tack]

If you find a breath controller exhausting, you should write more rests! ;-)

Of course that's actually the real value of a breath controller for winds and brass. I meant when using it for strings or an ensemble like with Divisimate.

[Luciano Storti]

Awesome overview, Matthias, thank you.

One other aspect touched upon by the creators is the ability to reassign parts played in with an ensemble patch to single instruments. This requires internal re-routing of course, but it does seem like a hugely useful and time saving feature when on a deadline. I’m far less enthusiastic about everything else, to the point that I’ll stick with the current, freely available Divisi Kontakt Multi script available now.

[Thomas Mavian]

Somewhat of a side-step but thanks to this I actually just got an idea of how I can (sort of) mimic Logics Environment mapping in Studio One via routings in WebMIDI and finally get something that can send to two (or three, four, whatever) different Studio One tracks simultaneously!

[Guy Rowland]

The more I've thought about this, the more it comes down to trying to re-create their strings set up, likely without the app at all. I do like the Symphobia Multis and do use them from time to time - excellent for very quick and dirty turnaround scoring - but mostly (and especially with strings) if I'm doing something properly I always want to write instrument by instrument, line by line. I think I write better that way, basically. In which case.... what purpose does the app serve me? I'd likely set up specific instances in VE Pro where all the routing is done there, and I can just play a single track in Cubase.

The main negative by some way is my terror of the sheer number of controllers in those Audio Modelling things. Can you set them up to just use two or three and it sound ok does anyone know? I guess it must be to get the results they did?

[ZeeCount]

If you are going down the modeling route, Aaron Venture's libraries are better if you don't want to delve too deep into controller editing, You can play his libraries with a single CC, and he has done a pretty good job of the spatilisation via his own custom IRs. The downside of his libraries is that you have less control, and thus are locked into his idea of how he wants the instruments to sound. On the plus side he is very open to criticism, listens to feedback, and is very active in updating his instruments. He has strings coming next year, which if they follow the same development path as his other libraries will start out ok, but with each iteration get better and better.

In terms of strings, I much prefer Sample Modeling's strings over the Audio Modeling ones. There are some really nasty resonances in the AM strings that get introduced with the vibrato that I've never been able to tame. The attacks in the AM strings are also quite problematic, and often leave you feeling like you are listening to a Saxophone rather than a stringed instrument.

I can make a comparison of the strings sound they were using in that demo with AM vs SM if you want Guy.