The next two frontiers in libraries

JEPA

Senior Member
are we needing soon to switch from Kontakt to another system? Would it be safe to back up the libraries in HISE, EXS24 or Sforzando? (sorry for deviating this thread, I need a short answer, thx)
...maybe the "next frontier" in libraries would be a multi-platform library system...
 

DSmolken

Senior Member
If you're concerned with looooong-term backup, as in decades, SFZ isn't a bad format - Sforzando isn't really being very actively developed by Plogue and I don't know about LinuxSampler much, but SFZ itself is an open and non-monolithic format, so it's relatively easy for future samplers or conversion tools to deal with.
 

JEPA

Senior Member
If you're concerned with looooong-term backup, as in decades, SFZ isn't a bad format - Sforzando isn't really being very actively developed by Plogue and I don't know about LinuxSampler much, but SFZ itself is an open and non-monolithic format, so it's relatively easy for future samplers or conversion tools to deal with.
thank you very much!
 
OP
justthere

justthere

Member
Lindon, thanks for the detailed answer.

Well no its not a subject change when you think about it. "The state of VI's" (especially in the world of "libraries" -its in your thread title) is very much driven by the capabilities of the platforms available to developers. Kontakt has not offered developers meaningful (or arguably useful) additional functionality for several years, and thus as the dominant libraries platform it is stifling innovation, and it's not as if developers havent been asking for additional functionality for years.
I see that - if Kontakt won’t allow it you have to go elsewhere, definitely. It is a pity that they aren’t pushing that further - but I’m trying to imagine how they would do that with the above features we have been throwing out there. They have to figure out what needs doing, and create the new engine that would accommodate that, including presumably a new programming/scripting environment.

Clever programming == innovation

When even "clever programming" isnt available to me, because the platform offers no way to program some new(or old as you point out elsewhere) synthesis approach, then the platform is stifling innovation. I would argue the very reason you see little innovation from sample libraries is because of there is so little room to maneuver on the dominant platform(Kontakt) - The nightmare we have to go through to offer legato and release samples in libraries is frankly edging towards the "too much effort to do" end of the scale. The programming environment (KSP) available in Kontakt is mindbogglingly old and out of date. Now thats fine, as it was never intended to be used even in the way it is used by most developers these days, but NI have made only cursory efforts towards keeping it up to date, and nearly nothing to aid in offering new functionality in libraries.

As I say developers want to offer new an interesting functionality - and it would be unlikely that any of us would be openly talking here about our specific ideas for future libraries - but we wont be able to do it on Kontakt, so (I predict) we will move to platforms that do support these ideas, and take our customer base with us.
Well, if that’s what it takes. I hope that someone comes up with a more versatile toolkit soon.

HISE is one such environment - its approaching V3, with a truck load of innovative enabling technologies - in the same time frame Kontakt moved from V5.3 to 6.1 - see if you can name a significant innovation enabler added by NI in that release history....I surely cant and I work with it every day.
That is impressive, but it’s hard to imagine people abandoning their DAWs to go live in the Linux hinterlands. I know it’s powerful, I know it’s efficient, I know it’s fast, but nothing I live with works on it. And nobody I know uses it. And HISE doesn’t have a ton of developers and users asking things of it and I suspect is light on its feet because it’s small - but Stanford and Yamaha own the interesting stuff in it (and how long has that been around?), and they are unlikely to throw millions at AI-driven modeling of instrument behavior. Again, I could be utterly wrong, but I feel like this area needs to borrow from other fields where this kind of work has been done. But what do I know? I only go as deep as programming rules in Mind Control and writing some scripts in Logic in between writing music.
 

aaronventure

Senior Member
Next level would be if it could analyze other aspects of the performance, and adjust more parameters than just the articulation. Kind of similar to what NotePerformer can do. It's a pity that NotePerformer's technology does not seem to have caught the attention of any daw developer yet.
Something that an AI analysis will never get right (not at the moment, at least) is human performance and expression. The "why" of it. Why play it exactly that way, when you could have played it a thousand different ways. The answer to that is something only a human can give you, and every one will be slightly different even if the overall message is mostly the same. But everybody will answer that "why" with a slightly different tone, volume and speed.

If you are going to sample instruments conventionally with separated articulations, you'll never get enough to be able to fit a recorded performance for every expression and every line.

What you're looking for already exists. And has for centuries. If you put written notes in front of an experienced player, they'll analyze it in real time and play it right out on their instrument. If your writing is good and your notation is clear, drawing upon their years of experience and reading scores, they'll probably get very close to what you want in the first couple passes.

If you know how to play an instrument yourself, you don't need to write down any notes. You just play it out of your head.

Since the whole idea is using virtual instruments ourselves, this is what you should look for instead. Look for virtual instruments that will let you truly express yourself. Then you can be churning out tunes in real-time. No AI will ever get it right for you.
 
OP
justthere

justthere

Member
You may be misinterpreting what I think the AI would be used for. I don’t want it to play it for me, and I don’t want it to make decisions for me. What I want it to do is figure out a logic for simulating ensemble movement and transitions based on lots of observation. And by the way, turn off your computer, because you already have one and it’s called the brain. And turn off your phone because you already have a long-distance form of communication called shouting. :) Just kidding. To me what the goal is is to be the first chair and the conductor and have the VI be everyone else and follow you in the way that humans do. There will still be a why, it’s just yours.

Something that an AI analysis will never get right (not at the moment, at least) is human performance and expression. The "why" of it. Why play it exactly that way, when you could have played it a thousand different ways. The answer to that is something only a human can give you, and every one will be slightly different even if the overall message is mostly the same. But everybody will answer that "why" with a slightly different tone, volume and speed.

If you are going to sample instruments conventionally with separated articulations, you'll never get enough to be able to fit a recorded performance for every expression and every line.

What you're looking for already exists. And has for centuries. If you put written notes in front of an experienced player, they'll analyze it in real time and play it right out on their instrument. If your writing is good and your notation is clear, drawing upon their years of experience and reading scores, they'll probably get very close to what you want in the first couple passes.

If you know how to play an instrument yourself, you don't need to write down any notes. You just play it out of your head.

Since the whole idea is using virtual instruments ourselves, this is what you should look for instead. Look for virtual instruments that will let you truly express yourself. Then you can be churning out tunes in real-time. No AI will ever get it right for you.
 

MaxOctane

Active Member
Not a technical advancement, but I've often thought a fun concept would be orchestral libraries tailored around specific pieces. Imagine a library around Dvorak's No. 9 mvmt II ("Largo"), and it has all the instruments and articulations to mock up the movement *exactly* -- the opening brass, the english horn, even the soft lyrical string quartet that pops up in the middle, etc. Maybe for ultra tricky passages, you even get a patch of brief phrases.

I feel a company could make a kick-ass product line with half-dozen thematic libs like that.
 

Lindon

VST/AU Developer
That is impressive, but it’s hard to imagine people abandoning their DAWs to go live in the Linux hinterlands. I know it’s powerful, I know it’s efficient, I know it’s fast, but nothing I live with works on it. And nobody I know uses it. And HISE doesn’t have a ton of developers and users asking things of it and I suspect is light on its feet because it’s small - but Stanford and Yamaha own the interesting stuff in it (and how long has that been around?), and they are unlikely to throw millions at AI-driven modeling of instrument behavior. Again, I could be utterly wrong, but I feel like this area needs to borrow from other fields where this kind of work has been done. But what do I know? I only go as deep as programming rules in Mind Control and writing some scripts in Logic in between writing music.
err, wow this is (as Dave has partly pointed out) way off the beam..As Dave says HISE is platform agnostic - and even if it wasnt it builds *native* VST/AU/AAX plugins. No one who is a user of libraries would need to even make a move away from their current operating system.

"and nobody I know uses it" - this I've come to call the "Mario argument" - hi Mario - which can be restated as "there arn't many products in it.." I got bored repeatedly pointing at the products that were already deployed that were built in it.

The next argument was "its not fast enough" - which if you spend some time in the KSP part of this forum you will see was also proven wrong.

If you went to the HISE site and looked thru the forum you would see there are "a ton of developers" asking for things, and getting them too!

But this is turning into another HISE thread so I will stop with this. It's just one alternative. Given the recent layoffs at NI I suspect my prediction of the demise of Kontakt (and several other NI products - apparently Machine is one) will come about before even I suspected.

Time to jump ship - or at least take a look where the lifeboats are stored...
 

halfwalk

Active Member
Time to jump ship - or at least take a look where the lifeboats are stored...
Conjecture? It's not like Apple stopped making iPhones after they laid off 300 people. Do you have a source that says NI is abandoning Kontakt?
 

JEPA

Senior Member
please name a NON PROPRIETARY Sampler Engine! This would be a good start for us all...
 

DSmolken

Senior Member
LinuxSampler is open-source, and more cross-platform than the name implies. Seems to support SFZ to a decent standard, though I've never used it and it doesn't seem very popular, so I've never gotten beyond installing it and failing to open one SFZ file. Whether it's based on something like JUCE, I don't know, but it might be as close to a non-proprietary sampler as it gets today.
 

TigerTheFrog

Froganthropist
T
Well NI has been laying off over 200 employees apparently... doesn't bode well.

This just happened a few days ago. I think it's very disturbing news and deserves its own thread.

Reading the comments from the staffers makes me seriously worry about the future of Native Instruments. It sounds like they are being Gibsoned. And I remember all too well what happened to Sonar.
 

Lindon

VST/AU Developer
Conjecture? It's not like Apple stopped making iPhones after they laid off 300 people. Do you have a source that says NI is abandoning Kontakt?
no of course I dont.. as NI have not (and probably never will) announce such a move. And yes of course it's conjecture, try reading their (somewhat vague) statement that covers their proposed "One Native" approach - and see if you can divine where Kontakt fits in that.

Meanwhile 300 Apple employees out of 123,000 = 0.24%
Whereas 100 NI employees = @20%

-- bit of a difference.

CDM has a nice coverage:


and


One thing to keep in mind - there hasn't for a long time been a dedicated Kontakt engineering team (as far as I know) - so cuts to engineering, UI and UX affect ALL products including Kontakt - which as I and others have pointed out in this thread hasnt had much love anyway in recent years.

Is Kontakt "going away soon"? - very very unlikely, if ever (Again conjecture YMMV)
Will Kontakt get the love its 3rd-party developers and users have been asking for? - Hmm, cant see how it "fits" in this "platform strategy" so (conjecture) - probably not.
 
Last edited:

Lindon

VST/AU Developer
Theres also this thread if you missed it:

 

TigerTheFrog

Froganthropist
I think that NI's official explanations should be looked at with a reasonable degree of skepticism.

Firing people is about saving money. "Reorganization" is public relations talk. You don't need to fire people to reorganize or repurpose them.

And these actions are accompanied by unhappy posts from people who work at NI who say that management doesn't understand the business.
 
OP
justthere

justthere

Member
Lindon, thanks for your insight from the trenches. Seems like you are saying that people have built many instruments on the platform even if maybe they don’t say it’s on the platform. A cursory google search yielded maybe two instruments like that, but I’m clearly looking in the wrong places. Can you tell me if any of the major orchestral players are using it and saying it’s their own engine? (Spitfire, OT, VSL, EastWest, and so on) Since we are talking about sophisticated handling of instruments along those lines, I’d like to hear about HISE in action. Their forum is full of requests and so on but I can’t determine who are pro developers and who are hobbyists. (I don’t know the players at all.) For example, the as-yet-unmaterialized OT player is supposed to handle legatos and releases in some new ways, if one can make it through their “keynote”. Wondering who created it.

err, wow this is (as Dave has partly pointed out) way off the beam..As Dave says HISE is platform agnostic - and even if it wasnt it builds *native* VST/AU/AAX plugins. No one who is a user of libraries would need to even make a move away from their current operating system.

"and nobody I know uses it" - this I've come to call the "Mario argument" - hi Mario - which can be restated as "there arn't many products in it.." I got bored repeatedly pointing at the products that were already deployed that were built in it.

The next argument was "its not fast enough" - which if you spend some time in the KSP part of this forum you will see was also proven wrong.

If you went to the HISE site and looked thru the forum you would see there are "a ton of developers" asking for things, and getting them too!

But this is turning into another HISE thread so I will stop with this. It's just one alternative. Given the recent layoffs at NI I suspect my prediction of the demise of Kontakt (and several other NI products - apparently Machine is one) will come about before even I suspected.

Time to jump ship - or at least take a look where the lifeboats are stored...