NotePerformer 5 is now available

You’re right, and I had completely forgotten! In a previous post, I had requested NPPE+BBCSO and NPPE+HOOPUS. @sjanssens sent me NPPE+Prime.

For this piece, between these three and straight NP, I really prefer NPPE+BBCSO. For other pieces with other styles, it may be different.

Dan,
With my setup and NPPE 4.5, I couldn’t use a full BBC SO orchestra; it was too overwhelming for my system setup. So, for me, Synchron PRIME is my go to out of necessity. Fortunately, it sounds quite good :slightly_smiling_face:

this refers to the number of instrument preloads that can be generated concurrently, launching a new process for each instrument.

Thank you!

the word “lossy” actually comes from Arne himself. It doesn’t mean it’s really audible, just allows a lot smaller file size.

I’ve mentioned this before, but it’s a great thing Arne has done with NPPE 5. Even though the library sounds are not as ‘full’ as in the original VSTs, it’s a fair trade off for me to finally be able to use the entirety of my $$$$ BBCSO and SSO 2024 libraries in a Dorico project. Having to ‘twiddle’ with them a bit is a small price to pay-for me. And, I still have my Synchron PRIME and Dorico templates for the ‘full’ sound in Dorico :slightly_smiling_face:

My dear friend composer, here is a link to my first work composed and released on social media using NPPE 5 with Spitfire SSO and MirPro 3D. This piece for orchestra is mainly “p-mf” with mostly strings. This link is not to promote my music but to illustrate the best I get so far with NPP5, so I hope it will be useful to help you to figure out how good is NPP5. From my point of view, NPP5 delivered.

Honestly? I really like the musical ideas, but I do not like the sound.

To my ear it sounds very artificial - especially in the middle register. (but, hey, I’m no expert. My audio setup is not top of the range and my physical ears are ancient. So perhaps disregard)

Listened through my Bose headphones. Quite decent sound. You can hear, as I discovered in my own orchestral works, that the Tubular Bells are out of tune. I find this true with my BBC SO Core, SSO2024, PRIME and other libraries in NPPE 5. I default to NP’s built-in sound for that. Sometime4s timpani rolls are out of tune as well. But I’m now able to use full instrumentation of these libraries, which I could not due before NP5- even in NPPE 4.5 because of the demands they placed on my system resources. As Janus mentioned, nice musical ideas.

No, I doesn’t mean to bother with CC programming ourselves directly. I am comparing with just using the orchestral VST with expression maps (and percussion maps) in Dorico without NPPE. I used to use VSL libraries with the expression maps provided by VSL, and EWHO Opus with self-made expression maps. They yield results much better than NPPE 5. I also don’t have to control vibrato speed, bow pressure, etc. However, Spitfire libraries do not work with Dorico directly for me. It must rely on NPPE (perhaps VEP might also work, I didn’t try).

To be short, my point is that the outcome quality of the different combinations are:

VST + NPPE 4 > VST + Expression Maps > VST + NPPE 5

From hearing your recording, adding MIR PRO 3D does not change my ranking. Trading off samples (e.g. round robin, legato programming, etc.) with NP’s humanized playback is just not worth and cannot be compensated by MIR.

Besides, I read Wallander’s Facebook post. I understand what he complains for the VST - as I also can’t use Spitfire libraries alone. I think he did a great job to solve the VST problems. This is why I switch to NP, and I don’t mind to pay more for such nice solution. Really feeling sorry to hear he could not sustain to solve the problem anymore. We perhaps need a black knight to work on a sustainable and quality-oriented solution to save us.

Thank you for your feedback. I agree to your ranking in terms of sound. But in terms of realism compared to actual orchestra performance I had the luck to have during concert, my template with expressions map I used with no CC programming could not be trusted. That’s why I switched to noteperfomer 4 + NPPE to avoid issues during rehearsals . Did you get a chance to compare your VSL + expression map with an actual performance of one of your work ? I am very interested by this solution.

I wonder why the aforementioned Wallander’s Facebook post isn’t showing anymore :thinking:

So here are a bunch of versions for comparison. You’ll need to adjust levels for each one to try to equalize them for fair comparison, because the loudness different for each one.

First, NP5 without NPPE as a reference: https://www.dropbox.com/scl/fi/bechvye9lmo9umk76pjve/And-Can-It-Be-NP5.mp3?rlkey=rwexdem1n9nsgepj6ybapl4u8&dl=0

Next, NP5 with HOOPUS (as another reference): https://www.dropbox.com/scl/fi/x1anq20kif3guej1lu6l8/And-Can-It-Be-NP5-HOOPUS.mp3?rlkey=7yz9b5lcfv48v08b36xsvw5d6&dl=0

Now, NP5 with Synchron Prime: https://www.dropbox.com/scl/fi/o78voec7kw27ezy0ryfbh/And-Can-It-Be-NP5-Synchron-Prime.mp3?rlkey=dn5bylhv30vta7rk0pg5xdhye&dl=0

And, NP5 with Synchron Full Libraries: https://www.dropbox.com/scl/fi/dxvroiez3y1eh2om194s9/And-Can-It-Be-NP5-Synchron-Full.mp3?rlkey=hw61q477abbqxrr6anpwbn72i&dl=0

Now for the versions using Dorico Expression maps without NotePerformer involved at all. These versions have no customization except where otherwise shown - just loading the playback template and hitting play.

First of all, to show some of the problems when you try using a poorly balanced, inconsistent and non-agile library (albeit one that actually has really great sounds underneath), HOOPUS with Expression Maps: https://www.dropbox.com/scl/fi/uefsb5w76atybzw8sqj2u/And-Can-It-Be-ExMaps-HOOPUS.mp3?rlkey=4i6vfh5k9xmk40u5e4jqukht5&dl=0

There’s a lot wrong with HOOPUS with expression maps version above. It gets worse with HOOPUS when you use even more articulations like col legno and more pizzicato and such because those are not balanced with the rest either. You can hear the sounds are great underneath, but the end result just sounds not great because of these issues.

Now, Synchron Prime driven directly by Dorico: https://www.dropbox.com/scl/fi/0xvu69wm20224uto76wcu/And-Can-It-Be-ExMaps-Synchron-Prime.mp3?rlkey=edye1diue657q0zwq05c5fmdu&dl=0

Better balance, fewer weird things sticking out, but still some weirdness here and there. Mostly it is a problem with agility since this piece is quite active with lots of moving parts and so you end up needing a good agile legato option so that things like the violin ostinato at the opening doesn’t get lost.

Compare with Synchron Full Libraries driven by Dorico (which have agile legato): https://www.dropbox.com/scl/fi/cmarirj31xhhorlwoqho3/And-Can-It-Be-ExMaps-Synchron-Full.mp3?rlkey=wkko8lj25wjns7eupogidmqkq&dl=0

Certainly the best of the Dorico Expression map options I would say in terms of balance and everything fitting together. Dorico’s playback isn’t smart enough to automatically interpret the correct articulations for parts of this, which is why we end up needing the agile legato more to hear some of the fast moving parts.

BUT.. fixing this does not take long at all. Here is the opening of the Synchron Prime version with expression maps with just a couple hidden playing techniques added to the strings at the beginning (took < 30 seconds to enter): https://www.dropbox.com/scl/fi/rkxe4lu8aqbo53ky2ihmp/And-Can-It-Be-ExMaps-Synchron-Prime-Opening-Only-With-Hidden-Playing-Techniques.mp3?rlkey=ixr847ra21riq49e71r80thna&dl=0

What was changed? See image:

  1. Existing “off the string” playing technique was set to trigger Natural playback technique, changed it to trigger Spiccato playback technique instead.
  2. Added new Playing technique called “Accent” to the list of playing techniques, to trigger the pre-existing Accent playback technique, and added it to the notes at the end of bar 2 to prevent them from being too short (otherwise they would be spiccato like the rest). Since this playback technique is type “attribute” it only affects this note.
  3. Hidden "ord."s added to return it back to the beginning, otherwise the whole notes in bar 5 would be spiccato as well which is certainly unwanted.

Those changes only took a short time. I could have added more shaping too by opening the CC lanes and quickly done a bit of stuff there. In my experience it doesn’t take all that much to get it up to and beyond the NPPE version by adding hidden playing techniques and CC lane overrides. But that’s only because the library is already balanced pretty good - with something like HOOPUS it would take a lot more work just to get it sounding reasonable.

Is it worth it to do this? The NP5 playback, especially with Synchron Full, I think is not bad at all really. It is probably the best of the above. But that doesn’t give you any control. Whereas with the Synchron driven directly by Expression Maps, you retain full control and can override articulations and add your own custom shaping and whatever is needed. It all depends on how polished the mockup needs to be.

It’s not really that much more work–simply dialing in the correct min/max in the volume dynamic section of the expression map usually handles the odd articulations. I’ll admit, though, that dealing directly with HOOPUS and Dorico expression maps does take considerable time upfront but that time is time I, personally, enjoy.

Yes. When VSL + expression map are used, dynamics, especially plain long notes, sound boring. This is where NP shines. Note durations are also very different to actual performance, unless you have the time to adjust one by one. If the piece is not too long, I might change some manually. NP will help you to change all according to its guess, but that does not fully eliminate the needs of changing note lengths. Fast running notes on string ensembles in actual performance sound soft - VSL too, but not HOOPUS or others. Balancing is also very different (unless MIR is employed), but the conductor (or you) can change it to imitate your audio demo or what you want.

If realism is needed, I would say NP5 will be closer after current known issues are fixed.

as usual, a sensible analysis. My take is that VSL excel in consistency of programming and instruments at an individual level but for some reason when everything is put together, the results more often than not sound rather sterile for emotional music like mine, (though this may be beginning to change). For this reason, I never use VSL for full orchestra. I’m not sure here whether it makes much difference if you use your own maps (or use the VSL ones) or NotePerfomer as I found when testing Synchron Prime in NPPE4 there wasn’t much to choose. However with both the BBC and Cinematic Studio, NPPE4 was a huge step forward and the latter is overall my favourite orchestral library. The timing issues which plagued both libraries were reduced, though not eliminated by NPPE4. The CSS trills issue (which I bypassed running natively by simply creating a hidden staff showing both notes of the trill) With NPPE5, they are almost gone altogether but here we have some new issues which have already been discussed. I was never convinced by Orchestral Tools Berklee, though that company produce some very interesting individual libraries.

I suspect that the best deal on that front nowadays is the Spitfire Symphony, though of course many will not have the resources to run it with NPE4 and initial reports under NPPE5 have been lukewarm at best.

I find the result really really good. Nice work. Also your question about a real orchestra is valid. Here in German I can move to a John Williams concert like every week somewhere and in all what I have seen and heard is. That often I have the feeling that many people never have listened to a real orchestra and how it sounds. Just my impression especially how much reverb a true hall has. Sometimes over 3 seconds. I don’t speak about a recording stage which is completely different in sound. So the realism you have created is really really good, even when you need to add a bit more balance. In my opinion. What I have learned from this entire forum thread is that there are people who want to use np5 the entire time but blame Noteperformer the entire time. Without trying or providing solutions. They just complain. But you have shown a way to make it better. Which is good in this entire debate here.

So don’t get too much distracted. This was a really good result with Mir3D Pro even when not perfect.
In the general media world. NPPE with Dorico and even stock Noteperformer 5 will provide enough quality for 99% of the jobs. Especially when you work for Influencers and social media where the big money is for composers and media people these days.
Noteperformer will improve there is no doubt about it. Many people will migrate to it within the next two years. ..
In Hamburg at the OMR aka the unofficial AI conference a week ago we have seen where the development goes and from what I understood is that we are only months away maybe less than one and a half year from a AI Cloud Sound Engine similar to what Arne said in this thread. ..maybe from Adobe or Splice or any other company. That will provide a good enough quality to battle any stage recording and any sample based Orchestral Library. From an economical standpoint the Game is over for sample based Libraries and Noteperformer delivers really well already, even when not perfect yet.

Absolutely on point thanks for the reminder. Especially spitfire SO 2024 with recent updates has really become great. Also the strings from the recording are really good on the iPad within staffpads own Noteperformer like engine, good enough for a really good travel sketch.

Thank you very much for sharing your experience. This is a very valuable piece of information.

Tough new developments.