evidence for this is that the trill slows with your rit…
And those notes ramp up in volume…
Then the longer the note sounds, the louder it gets due to the swell on the start of the note.
Therefore the slower the trill, the longer the note, the louder it gets. You said your trills got louder as they slowed.
I see this issue all the time where if I have a long (e.g. whole) note followed by a bunch of short (e.g. 1/16 or even 1/8th) legato notes, there is a big drop in dynamic in the slurred shorter notes…
In some tempi, you get a massive whip-tail on the end of each note as it hits the biggest part of the swell before the note changes. It’s pretty awful.
OK – I’m with you, I think. The thing is, of course, that a sampled trill is not built up of individual notes in the sense I think you mean – it is only when using the NPPE technology that this occurs as otherwise, the trill stays at a constant speed as it is a single sample. I assume you have used CSS natively as well? So the question remains —how can we deal with the behaviour in NPPE.
It’s up to Dorico whether it simulates trills with a stream of notes. So I suspect this is where it’s coming from. Especially as you said there’s no trill articulation in the articulation map.
If a notation program has no trill articulation or know of any way to access a trill patch, it would try to emulate it.
Arne mentioned that NP has some way to see this and convert it to a trill articulation / patch. But it would have to watch the stream of notes for a while to see if it’s a trill. So may depend on tempo of a measured trill. if it was using the trill patch then I think the trill speed wouldn’t change.
yes, OK - obviously NPPE does work in a different way than normal. Arne says it uses the trill sample but yet it doesn’t use the trill articulation (as can be verified in the Key Editor) which is why it needs to find a way to watch the stream and look out for trills as you say.
To me what is happening is that the trill starts off using the sample and then when the rit, kicks in, it suddenly switches to the emulation which is why the result is uneven with a very clear (and undesirable) transition
Ah that makes sense. As the stream of notes from Dorico slows, NP stops recognising it as a trill and passes the notes through instead… So I’d expect the trill to stay same speed for a while, then cut over to much slower and louder, and from there slow down further.
exactly! So perhaps Arne might want to ponder this and see if there are some tips to improve performance in this situation. As I said before, varying the trill speed in the Properties hasn’t so far helped. Ideally, there should be an option where the sampled trill simply continues if this could be implemented.
Correct. The libraries that do work well with that kind of “macro for mixing — pan and depth” have monoish dry sounds. Like HSO. Adding reflections and reverberation to already “in a room” or “in a hall” instruments will add muddy sound… Not really what you are after
It may be able to do that with the built-in NP sounds.
But the samples in the sample libraries have baked in panning, and it ties into the room sound (since by and large they are wet) so can’t really be altered well.
NPPE does panning for ambient sounds, but it’s a complicated process where we must compensate for the baked-in panning and do careful processing on a per-microphone basis. You need information about the underlying sound to do panning effectively, which is why we do it internally. You must also pan prior to adding reverb.
I’m searching for solo strings to use with NPPE, and have my eyes on Cinestrings Solo. I tested some individual articulations, available via Musio, but that made me unsure… does anyone know if those available via musio are flattened (less dynamic layers) or identical to the full version? TIA
If you’re not in a hurry, I understood that Spitfire’s Solo Strings were to be NPPE’d eventually. I hope so, because I already own them and use them, and I see the great value Arne’s magic could bring there!
@Wallander, in the attached excerpt of the first few bars of Debussy’s Pelléas et Mélisande the cellos in bars 1 & 2 and in 8 & 9 seem to scoop up in pitch when using NP4/NPPE with Synchron Prime but sound fine using NP4 alone or Dorico’s default HSSE/HSO sounds. Also, the cello notes in bars 10 & 11 are subject to the same scooping when they play but often no strings sound in the first two beats of bar 10. These issues seem to be limited to playback using NP4/NPPE with Synchron Prime with which I am mostly very happy. Are these known problems in the initial release of NPPE that will be fixed in an update? In any event, is there anything I can do now?
EDIT: The notes also play back fine using the Synchron Prime playback template so the issue seems to be related to NPPE.
Yes, i have them too, and hope they enjoy a high priority for NPPE’ing… they are wonderful. I need to finish my 3rd string quartet before the summer and could use something to boost the inspiration…hence the hurry
Note for @Wallander which is the automatic setup. It works great! When it works. Which is when it’s the default and is used for all your instruments. Whenever I mix and match (use NP and some other VST’s) then NotePerformer stays blank and doesn’t sync up. Normally to fix this I’ll do the clumsy set Playback Template to Silence, then back to NP, but obviously that won’t work here. Usually to fix I have to carefully start over by having NP be default, then carefully switching over those that need to be another VST.
So given that now the playback quality is so high I think people will be doing more mix-and-match of VST’s with NP - if you could please make this easier, or work better, like giving us the ability to configure NP without it trying to guess - or something - that would be really helpful.
Thanks - I’m tripping over this problem more frequently with 4.
You can make a playback template that combines your other VST and Noteperformer. Set instrument overrides for those instruments you won’t use from NP, and NP will just fill in the remaining…
I’m still puzzled with how NPPE handles legato in combination with BBC SO core. Legatos are non workable for me because they cause undesirable rythmic changes.
Here is the playback with NP, to me sounding rythmically “accurate”:
Here is the playback with NPPE + BBC SO core.
The pairs of slurred eighths feel not like equal eighths but almost like consisting of three parts instead of two: 1 sixteenth (first note) + 1 sixteenth (second note) + 1 eighth (second note) or something of that order.
To make sure it’s not my brain fooling me I’ve slowed down bar 3 beat 3 to 25% of speed and added a click on every sixtheenth beat.
NP:
Sounds “natural”
NPPE + BBC SO core:
Sounds extremely “hurried” (something a jazz improviser would do to anticipate)
With some basic frequency analysis in Audactity I find that the Ab component indeed already starts to be showing up in the time zone of the second sixteenth note (the white area is approximately the zone where the Ab is detected):
Yeah that’s interesting @Wimuz , I hear it too. I didn’t read your post other than getting the idea ‘there’s a difference’ and then I listened to the two - my first thought was the NPPE+BBCSO felt like it was leaning into the beat, almost but not quite rushing. NP feels more poised and in control.
So in theory I wasn’t biased by your impression not having read it beforehand, going back and listening back and forth a few times I’m pretty sure it’s there. It would be very interesting to compare these to stock BBCSO - can you mock that up? Shouldn’t take much time.
