I don’t think the level differences should be quite so drastic. Something may be wrong here. Can you send this specific example over so I can have a look / try it out?
Duality is configured to play staccato a bit louder than legato but I don’t think the difference should be quite this pronounced. This is not an error but was on purpose on VSL’s part, as it has been raised with them before, and they indicated that it was done on purpose as performers will usually play staccato a bit louder than long notes even when marked at the same dynamic. It is possible in the player to rebalance the articulations - for instance all of the “short notes” could be made quieter by a certain number of dB. But before you get to that point I’d like to make sure it is working properly first.
I had reported the excessive difference of level between longs and staccatos in Duality Strings, but the answer was that staccato is always louder. But it is too much for me.
My guess is that they decided to default on the louder attitude staccatos have during ostinatos in modern film music.
Fixing it is easy: in Synchron Player select the Short notes cell, then lower the volume in the Edit pane:
I’ll certainly try what you suggest, Paolo, thank you for the suggestion.
I suspect CSSS with NPPE may be a better fit for me, as the legato phrasing in expressive string parts (out of the box) is much improved compared to Duality Strings. Here’s an example to illustrate the difference:
Update: Also, markings such as sul pont and con sordino result in no playback if the corresponding samples haven’t been purchased. While that’s perfectly understandable, it’s unfortunate that the VSL automated playback template creator doesn’t seem to take this into account.
It does look like it is doing that by default, I must have not written many passages that mix the two together - or at some dynamics this difference might be lessened.
Aside from doing this slider change to the level of the staccatos, another alternative (and may have better results) is to manually the velocities for the staccatos. This is actually quite easy and quick due to the way that VSL has set things up.
Velocity only controls the dynamic of short notes and the dynamic of long notes is controlled with CC1. So in cases like this it is possible to quickly drag your mouse across the velocity lane with the pencil tool selected:
So this is a manual fix, but it doesn’t take too long because you can just drag horizontally with the pencil tool and quickly fix everything. It doesn’t matter that the natural/legato notes get lowered too in velocity because it doesn’t affect them.
You could also make a custom playing technique like “quiet” and hide it, and make a new base technique in the map that matches staccato + quiet (as a duplicate of staccato) and reduce the max velocity there. It should scale to the lower range. This would have the same effect but would automate it. You would just have to add the word “quiet” overtop as a hidden playing technique. The alternative would be to make this adjustment to the main staccato one and add a “loud” variant that gives the full range for those moments where you want to get to the highest dynamic (because otherwise it will be capped at a certain point).
The reason the legato phrasing is not good here with Duality Strings is that the VSL maps are not configured to use the legato largo which is so much better than the original and more connected. I would recommend duplicating the original VSL map, reprogram the duplicate to use the legato largo instead, and assign that to the part.
Thank you for the suggestion, Michael. I do wonder whether VSL plans to update their playback template creator to account for this. I’ll reach out to them and report back…
Well, but are you comparing SDS played directly by Dorico, with CSSS played through NPPE. It is not comparable.
Sounds played directly from Dorico are only slightly ‘humanized’, and are mostly played ‘robotically’. The ones played through NPPE are performed by the sophisticate interpretation algorithms of NotePerformer. Computer vs. modeled human performer.
If you like, you can get a demo of Synchron Prime, and see how it works with NPPE. The strings are a bit different (being derived from Synchron Strings Pro), but you’ll get an idea of how the “VSL sound” works for you.
They don’t plan to at the moment, which is something I strongly disagreed with with their rep on the forum (you want the Dorico playback to show the library in the best light and the default legato is so much worse).
Just for some information, good legato requires some “lag” or “latency” in response, which means you won’t go to the new note instantly. CSS requires this for instance. The old default VSL legato aimed for zero-latency so that you would instantly transition to the new note. The new legato largo is more like CSS where there is lag before you switch notes before it gets to the new note (as it transitions). This extra time is the reason why it is so much better and so much more natural.
What has to happen when you’re dealing with this lag is the note onsets have to be moved slightly back in time so they are just before the beat. This can be done manually (by adjusting the start offset for the notes, which changes how the sound but not how they are notated) or automatically with Dorico’s feature of negative track delay per articulation (done through the expression map).
The only exception to this “lag” ends up being the first note of a legato phrase, because nothing comes before, there is no note to transition from, and so the lag is removed. Dorico unfortunately does not currently have a smart way of handling this, so if you apply a negative delay to “legato”, all of the notes will arrive at the correct timing except the first note of each phrase will be a hair too early and have to be manually adjusted. (CSS is the same way)
The reason VSL did not want to implement the legato largo in Dorico is because of this first note. They would rather that users get much worse legato in Dorico by default than the library is capable of, because otherwise users may have to manually adjust timings of initial notes in legato phrases.
To me, it is not that big of a problem to manually adjust the onset of the first note of a phrase when it is only the one note, and I think it would be better that users get the much better legato largo than the regular legato even though it might mean manually adjusting onsets. You could even make a new hidden technique of type “attribute” called like “First Note” that you could put over the first note and that could apply a different delay to just those initial notes of the phrase, to avoid having to edit start onset directly. But try to convince VSL of that…
You make a very fair point, Paolo. Perhaps my search for an NPPE replacement will never be fully realised if what I’m really hoping for is the same level of musical response from a playback template. Some libraries do read ahead and attempt to produce more authentic results, but I’m probably setting the bar too high.
VSL doesn’t have any sort of “magic” in it like NotePerformer - it relies entirely on the Dorico features to produce a musical response. As Steinberg adds more “auto shaping” etc into Dorico based on context, this will directly benefit such libraries.
Going away from NPPE you should expect things to not be 100% automatic and you will likely have to do some manual work to get good results. However in my experience it usually does not take very much manual work at all to get the results as good, and then it is only up from there, because you have full control over the result so you can get even better results than with NotePerformer.
Thanks for explaining this so clearly, it really helps put the issue into perspective. I can understand why VSL hesitated to implement legato largo in Dorico, given the complications with the first note of a phrase. That said, like you, I’d much rather have access to the more natural legato, even if it occasionally required manual adjustment. In expressive legato passagework, string players often begin a phrase just before the beat, much like a singer placing the “f,” “sw,” or “th” of a word before the beat so that the following vowel aligns with it, which makes the timing feel perfectly natural.
For my own work, though, my priority is to achieve the best possible playback with the least possible intervention. The prospect of manually tweaking staccato dynamics or adjusting note onsets is simply a step too far. In musical theatre, time is always at a premium and revisions are constant. Even modest interventions add up quickly, and documenting them so others understand what’s been done only increases the workload. For that reason, VSL isn’t the right fit for me at this time. Still, I’ve enjoyed exploring it today, even if only briefly. Thank you to you and Paolo for sharing your expertise.
Yeah, I would say NotePerformer is the only option that is really zero effort, at least as it stands right now.
Even if you look at something like MuseScore, which has some great sounds, it has balance issues, so you have to create all sorts of hidden things to improve the playback, or over-notate the score. So you still have to do manual intervention even there. NotePerformer does the best job of anything out there at replicating real orchestral balance and articulation selections.
The long term solution, and best solution I would say, would be for some of this “intelligence” to be transferred to Dorico itself, so that it knows how to make better default choices in cases where no markings are present, based on the musical context.
I spent a little time away from Dorico recently, having a series of Sibelius jobs, and just returned, only to find that NPPE 4.5.1 is very flaky with Dorico 6, pretty much to the point that it is no longer usable. I had been using the BBCSO Pro Sounds. Their solo string sounds made a huge difference in Pro over Core that lacked them. I upgraded my computer to run it and achieved some amazing results.
After spending all day yesterday battling with NPPE, thinking that it had to be a BBCSO issue, only to find out that it still worked perfectly with the latest version of Sibelius. Today, I attempted to use BBCSO Pro without NPPE to mixed success. NPPE compensates for some of the BBCSO balance issues, and for me to use it, I would have to change all my dynamic markings, especially on the low end. The soft dynamics are way too soft. I like the presence of the lower strings, yet the piece I was working lost the punchiness of the brass. NPPE gave me a more natural balance, although using straight BBCSO avoided some of the Noteperformer eccentricities, bumpy slurs and wrong articulations. Somehow, switching back and forth between Noteperformer and BBCSO also corrupted my file (won’t play anything now), so I now am going through the tedious process of reconstructing my score from a post mXML state.
I can’t attest for VSL Prime, since I don’t have it, but if I completely changed the dynamics of my orchestral scores, I could replicate the more natural sounds of NPPE. That said, I want to be able to send my scores off with the correct dynamics in them. I don’t want to have one for live performance and one for a demo.
Basically, my new working method is going to have to be to create the score in Sibelius, record the demo with NPPE/BBCSO, and then export into Dorico 6 to clean up and extract parts - a convoluted way of working. Alternatively, I could go back and work in Dorico 5, since NPPE is now dead. Unfortuately, I can’t have NP4 and NP5 on the same system.
FWIW I’m using BBCSO Core (not Pro) with Dorico 6 and NPPE 4.5.1, and it’s been working fine, both with musicxml files and files created in Dorico. But I haven’t tried BBCSO Pro yet….
It’s a lot more money, but the solo strings in Pro are worth having, much better than NotePerformer will give you. You’ve heard the demos of my string quartets (1 and 3). Those are Pro’s solo strings. I haven’t really ventured into all the other exotic instruments they provide, but it is odd that they don’t include saxophones or alto flutes, but they do have cimbassos and contrabass clarinet.
Following up on my earlier posts: for those interested in working with Duality Strings in Dorico, I’ve been in touch with VSL about achieving a more consistent legato in slow, expressive passages. For the example below, they suggested the following approach:
1.) I have added slurs from bar 87 to bar 95 so that playback doesn’t switch back and forth between normal long notes and legatos. You can hide such slurs, that are only relevant for playback, but not for layout, by making them transparent (Common/Color/Alpha channel: 0).
2.) I have reloaded the Synchron Player preset and made Legato largo available this way.
3.) To switch to legato I have added a B1 keyswitch to the score. To compare with normal legato, you can change it to an A#1 switch.
4.) I have raised the release time in the three Synchron Players from 64 to 82.
5.) I have turned down the cello in the Dorico mixer. Our instrument presets take care that the instruments have a natural volume balance. Unfortunately the Dorico mixer flattens these settings. So I recommend making some volume balancing in the Dorico mixer.
I think with those internal slurs Dorico is shortening the note at the end of each slur which is creating the “pause” in the line. I would suggest selecting just the internal slurs there and disabling the playback effect in the properties panel at the bottom so that only that outmost playback slur is having a playback effect.
The way Dorico handles these types of passages that are obviously meant to be highly legato but are instead a mixture of slurs and single unslurred notes is not very good at the moment and is one of the weakest parts of the playback engine. It is a difficult problem to solve of course, at least in a way that works for most passages of music and most libraries out there, so it is no wonder that they haven’t really tackled it yet. But it is one of those things that I hope they will handle in the future.
Disabling the internal slurs definitely improves the overall sense of flow, but I’m even more struck by the results from NPPE and CSSS.
Of course, it’s not a fair comparison, but it does underscore just how much further Dorico will need to evolve in order to reach that same level of musicality.
(Duality String, internal slurs removed)
(NPPE + CSSS)
I am now keen to explore CSSS direct from Dorico, which I guess would be fairer comparison with Duality Strings.
The NPPE+CSS is doing a fair bit of shaping that is not written in the score (I assume you mean CSS as CSSS is a solo string library, not section). This is another thing that Dorico is not doing at the moment. But in this situation, I do feel like the way human performers would shape the line is a bit different from how NPPE+CSS is doing it. In these cases I would add my own CC1 lane shaping - these few dynamic markings are insufficient to drive this and get very musical results, as it stands.
The CSS might have more vibrato too - you can add a hidden “vib.” mark at the beginning of the passage if you want VSL to use molto vibrato.
