I’m a very satisfied end-user of Vienna Symphonic Library and Spitfire Audio libraries, but I’ve been hearing more and more about Orchestral Tools, mostly about their Berlin Series and SINE player… Has anyone here had any experience with OT? And if so, how would you compare it to VSL and SA?
Another question, this time on NotePerformer’s AI system… Does anyone know of any system like NP’s, but that would allow the end-user to use his/her own library of choice? My guess would be that the answer is no, but who knows.
Any hints or insights on these issues would be most welcome.
It seems Orchestral Tools hasn’t much of a user base among doricoans… That’s unfortunate, it would be great to have some input on it from experienced OT end-users.
I have OT’s Berlin Woodwinds and Stings. I switched to the winds from VSL and the stings from Spitfire. I like both string libraries. It’s going to be tough to get me away from BWW. Haven’t used the SINE player yet as the libraries haven’t as yet been converted.
I’m a very satisfied end-user of Vienna Symphonic Library and Spitfire Audio libraries, but I’ve been hearing more and more about Orchestral Tools, mostly about their Berlin Series and SINE player… Has anyone here had any experience with OT? And if so, how would you compare it to VSL and SA?
I have experience with both SA and OT libraries. I found these of OT much more realistic, and better sampled than the ones of SA.
I’m using OT Berlin Series as main orchestral library and SA Symphonic Series as doubling. Unfortunately due to sampling process artificial organ-like sounds appear while using a single library, so that’s why doubling is necessary.
SINE Player looks interesting, but for now it doesn’t support the Berlin Series. But in the future it will, and surely this would make the sound even more realistic.
Another question, this time on NotePerformer’s AI system… Does anyone know of any system like NP’s, but that would allow the end-user to use his/her own library of choice? My guess would be that the answer is no, but who knows.
I think that there is no any sampler with AI capabilities out there, yet. Well, SINE has some AI features.
Thank you both, I guess I’ll wait for the Berlin Series’ packs to be ported to the SINE player, but from your replies I definitely get the feeling that it would be great to include Orchestral Tools in my tool set.
Thurisaz, on the AI issue, my guess is that AI like NP’s but open to other libraries will take a while to become available. I hope I’m wrong about this, though. We’ll see.
When one considers that different libraries have different keyswitches and other means of creating articulations and playing techniques, it seems highly unlikely that any generic AI would exist to cooperate with various, different sound libraries (without extensive user customization) unless companies come together to create a standard for keyswitches, ports, and the like.
Derrek, the need for “expression maps” is already a thing, as this forum can attest. And I think we actually agree that there would be many advantages to standardization.
So yes, of course it would also be an issue, but it wouldn’t be neither a novel or unsolvable one. What seems to me much harder to solve is the issue of access to proprietary audio samples in very large numbers, which would be needed if an AI is to be able to analyze and operate on sample libraries… Professional sample libraries not being Open Source, they aren’t open targets.
This actually seems to me to be the hardest problem facing any AI system like the one summarized in my original post.
My understanding is that NotePerformer is not just samples, but a blend of samples and synthetization. I think working out such a combination might be the stumbling block for any software that tries to treat 3rd party sample libraries as NP does its own.
If I have misunderstood how NP works, I’m sure someone will correct me.
L3B, every professional sample library (including, and of special interest to this thread, the libraries from Vienna Symphonic Library, Spitfire Audio, and Orchestral Tools) is a full-fledged sample-based Virtual Instrument. Either I’m misunderstanding what you meant by “synthetization”, or Noteperformer is definitely not exceptional in that regard.
Agreed, Dan, but not in terms of “synthetization” (L3B’s word of choice) as I understood it. More than anything else it’s the AI system that makes it exceptional, it seems to me, and that’s actually their banner.
L3B is correct. There are two ways of doing Virtual Instruments. Samples and Waveform Synthesis. Noteperformer blends the two.
In pure Waveform Synthesis, no samples are used. The waveform is generated purely by equations.
Thank you, Craig, for explaining what L3B actually meant by “synthetization.” I didn’t know NotePerformer relied on more than sample-based synthesis for its sounds… I wish I knew more about how they go about it, it should be a fun read. Anyway, to put it mildly, the individual sounds don’t actually sound better than VSL or SA, it just performs “out of the box” a better playthrough thanks to its AI system.
Noteperformer’s sounds don‘t use key switches. No separate staccato samples and long notes. They are playable like i.e. Samplemodeling, Audiomodeling or Aaron Venture instruments. They react to velocity, note length and a bunch of midi controllers to sculpt the different articulations. So the AI wouldn‘t work with standard sample libraries.
And it‘s even impossible to play something in the DAW with one sample library and replay it with another. It often even doesn’t work within a single sample library to copy midi from one track to the next without the need to edit CC data or velocity or timing to get musical results. What Arne did to make Noteperformer run is a kind of adjustment and calibration miracle!
So, Saxer, NotePerformer also relies on acoustic, physical modeling for its sound synthesis… I couldn’t find any detailed source on NP’s patented technologies, I’m under the impression that Wallander Instruments has “blackboxed” them. On the other hand, it’s clear to me that you’re overplaying it… It’s hardly anything miraculous, there’s no reason to suppose that any of the issues you raise would be impossible (or that hard) to handle by an AI system with an appropriate model, database of its target sample library.
The point is that NotePerformer’s AI is specifically tuned to the sound-manipulation capabilities of the NotePerformer sounds. Since most sound libraries use different codes for sound-manipulation specific to each company’s libraries, a generic AI engine is currently out of the question.
I use the Wallander Instruments a lot. I play them via wind controller or breath controller and they react directly to the playing. A complete different way sample libraries work. It‘s a difference to play a short note or to trigger a one shot sample recording of a short note. Playing the short midi note longer wouldn‘t extend the sample length but it works with Wallander Instruments. So yes, using AI might work for samples too if you create a complete different sample selection system. Different AI approach. But it‘s not a compatible one to Noteperformer. Though I could imagine that it could work with similar virtual instruments like Samplemodeling etc.
Saxer, the points you raise have also been adressed and answered before, including my previous reply to you. I’d just like to restate this a bit less mildly, in terms of individual sounds, Vienna Symphonic Library and Spitfire Audio (I’m sure this is also the case with Orchestral Tools) sound much better than NotePerformer, for me there is absolutely no contest there.
It’s obvious (to some people) that a $1000 library is going to sound ten times as good as a $100 library.
Just like a $1000 iPhone obviously makes phone calls better than a $100 Android.
Some of the demos on the VSL site itself sound completely unrealistic compared with any human ensemble. If that’s the best they can do to advertise their own products, they need to think again.
