Tried. continues to close 
EDIT: Suddenly it has reopened normally, for whatever.
Here’s some background. I understand that many will be surprised by this, but knowing all the details, I have 100% confidence in the path we went for.
With NotePerformer 3, we have already reached peak maturity for what a lightweight library can do, at least in my hands. We already made minor, incremental updates to them over many years. If you look around, there are almost no modeled options for strings or percussion, they don’t usually approach the state-of-the-art sound quality for the casual user, and they’re not necessarily lightweight either. Modeling high-quality strings and percussion is not within reach of current technology as I know it.
Deep-sampled libraries recorded in a natural space were our only viable alternative for dramatically increasing the quality beyond what we already had. The libraries we support with NPPE (NotePerformer Playback Engines) represent the state-of-the-art in deep-sampled libraries (there’s a lot more, but this is a subset). Experienced companies with the right skills and access to the right people and venues recorded them. We could never produce a deep-sampled library to match on the first attempt. Even if we did, NotePerformer 4 would be like this, but with just a single deep-sampled library to choose from that’s mandatory, require a powerful system with lots of RAM, and with a mandatory upgrade fee and a built-in reverberation that would be very disturbing for users who don’t want the natural ambiance. With the current solution, everyone stays on the latest technology for free, and the lightweight technology runs side-by-side with the optional third-party options.
There’s also a problem of sound fatigue. If we did create a new sound library, our users would quickly grow fatigued with that, too. It makes more sense to have a modular engine that can support many VST libraries recorded in different venues or produces sound with various technologies, which we can update with state-of-the-art sample library technology.
Another angle is that NPPE uses our factory NotePerformer sounds as real-time guides for the NPPE engines within the engine itself. We spent a decade tweaking those sounds for balance and articulation; I have very high confidence in the balance of NotePerformer 3 and wouldn’t want to gamble with changing them at such a critical point of development since it offsets the entire ecosystem. Even if we changed them, I’m not sure what to change, and I don’t know if it would be for the better, but there would be many opposing views. As was always the case whenever something in NotePerformer changed.
The YouTube demos of the playback engines represent the out-of-the-box sound using the Mix/Main perspective of the library and the built-in reverb (only a touch of reverb) because we don’t want to give the impression of a sound that requires customization beyond that of an ordinary user. We have managed multi-microphone support for applicable libraries, which is very powerful. There’s so much that can be done to customize the sound beyond our baseline renderings, which is the same process as working in a DAW, but much easier to do in NPPE.
There’s little change from the user’s perspective to the core software, but a lot changed under the hood. The support for native Apple M1 and VST3 for Dorico were significant NP4 updates which we released prematurely out of necessity. There were stability issues with OpenGL in Windows that only affected some users with specific drivers, so NP4 was equipped with a new Direct3D-based engine which is a significant change. These are all very high-effort updates with little return for most users. Moreover, NotePerformer has almost no known bugs, so there’s no natural reason to have a long list of fixes.
This introduction of NPPE is also a significant shift since it’s a natural position for supporting big band libraries and other styles requiring a separate playback-rule engine. Nothing prevents us from supporting modeled instruments by other developers, although they present unique challenges that must be resolved.
I must also be said what’s unique about our software was never the sound library but how we integrate it with notation. It makes sense for us to become an intermediate host that applies our technologies to sample libraries by other developers, as an asset to the ecosystem, rather than trying to dominate that space ourselves.
37 Likes
The only reason we didn’t is that SYNCHRON-ized SE requires onboard MIR to shine, and our engine can’t handle the CPU strain since we run dozens of instances for every instrument. We must do this to deal with legato polyphony and countless situations that can’t be effectively handled on a single MIDI/audio channel.
3 Likes
We do a ton of post-processing for every single note. We have dozens of plug-in instances for a single NPPE slot and manage timing, pitch, envelopes, and more advanced processing to guide the output of these libraries in the direction of NotePerformer. We’re also processing the plug-ins asynchronously, so there’s no 1:1 relationship between the time in the score and the time for the plug-in. We could never translate what we do to MIDI, as it would retain very few qualities.
5 Likes
I would think libraries like Sample Modeling and Aaron Venture’s Infinite series would be a perfect pairing with your playback engines. You mentioned some technical hurdles to overcome though but wondered if that’s something you’re actively pursuing? I think this has exciting possibilities.
2 Likes
The problem is that NPPE splits instruments over dozens of instances and combines them coherently from articulations. We can’t use performance patches effectively (you could say we create our own performance patches) meaning some of the benefits of modeling are lost in the process. The main obstacle, however, is that many of these products rely on onboard reverbs, and that’s not sustainable from a CPU point-of-view when you have several hundred instances. Chances are we need to output all sounds dry.
Also, whether the modeled products should be for orchestral or big-band use is not obvious. Personally, I think deep-sampled libraries have an edge with orchestral because the natural ambiance has a dimension that no reverb truly captures. Meanwhile, many modeled products have great qualities for orchestral but are even better for big band.
1 Like
Arne, are there any demos available that use NPPE?
Here are my demos using NP4 + CSS :
-
Rainbow Splitter (MK3):
[2023][Op.30] 七色乖离 第三版 (NotePerformer 4 + CSS 弦乐)_哔哩哔哩_bilibili // CSS configured Mic with boosted Close Mic. -
Celestial Battle (MK1):
[2023][Op.32] 天神共鬥 MK1 (NotePerformer 4 + CSS 弦乐)_哔哩哔哩_bilibili // CSS original Mix.
P.S.: Forget about the upload time (2021-10-06) since I just reuploaded and replaced the video file.
Hi Arne,
thank you for the incredible amount of work and labor you’ve committed to this software. I’m very excited, as I’m really looking forward to be able to use libraries with Sibelius and Dorico where no sound sets existed before. Technology is making great moves forward for those of us who enjoy working in notation programs.
Thank you Arne! 
3 Likes
In their youtube channel you have demos for each library:
1 Like
We have a series of out-of-the-box renderings with each supported library. These are unedited renderings with a baseline template, using the Main/Mix perspectives.
Perhaps naively I took the plunge straight away and purchased the Opus Hollywood Note Performer engine licence - mainly on the strength of trusting the intelligence that goes into every Note Performer release. The Opus player is already a complex and very layered piece of work - maybe if note performer hosting and translation simplifies things via the new sosund engine that will be good.
Retrospectively a bit suprised that if i want to licence another orchestral library I’d be paying another pop of $69 or $89 though - one would have thought additional libraries being added would be cheaper than this.
It sounds like there may be no sonic improvements to the Note Performer library itself - is this really the case?
Was also slightly disappointed when I did a quick test that the Note Performer instruments still seem to load themselves onto one channel of the Dorico mixer. Someone did a seperate outputs fix for this a while back I know but I would have thought NP4 would have also resolved this.
I have never found the NP mixer the most graphically or ergonomically pleasing aspect of the library anyway - more work clearly went into the Dorico mixer visually. So I am hoping with the new 3rd party NP sound engine I will have a more satisfying experience.
NPPE slots can be routed to different outputs of the first NotePerformer instance in Dorico. Unfortunately, the stereo output name won’t match, but it’s not under our control.
To get multiple outputs, I think you must reassign the Playback Template.
2 Likes
Sorry, I didn’t scroll right to the end.
For my money, the Hollywood and BBC ones sound the best.
(Well, it would have to be someone else’s money…)
Speaking of which, they both have sales on at the moment.
2 Likes
It’s free if you already have NP3! Just downloaded it myself.
2 Likes
That’s awfully nice of Wallander! I’ve got BBCSO and HSO so will try it shortly …
1 Like
You do have to pay for the engine licences though.
Jesper
Edit: which I think is fair.
4 Likes
Sorry, just woke up and I’m slow … I’ve got a licensed copy of NP 3, what engine are you talking about?
Sorry if I missed it elsewhere, but I don’t think anyone has linked to @davemacdo’s great review on Scoring Notes yet, which does a really good job of explaining all this, including how the NPPE licenses work.
7 Likes