Inconsistencies in the Expression Maps (VSL in Dorico Pro 5)

[DirkMM]

Download "Experiments in Scales _ VSL in Dorico.rar"

I am new to Dorico, but I have been using VSL for a considerable time with Cubase Pro. I decided to switch to Dorico because I want to write directly into the score. However, I also want to create the most accurate demo possible of my music. From what I understand, the Expression Maps should ensure that the nuances in the score result in the desired nuances in the sound — meaning that the notation and sound should correspond. But based on my experience so far, this is absolutely not the case! Here is what I have done.

I am using Dorico Pro 5, and I have downloaded the VSL Expression Maps. I set up a track with "Synchron Duality Strings 1st Violins – VelXF sus" and chose the "VSL SY Duality Strings – Violin, Violas" in the Expression Maps. I began composing music and noticed a completely different sound than I had in mind. So, I stopped writing and set up A VERY SIMPLE EXPERIMENT: I had these violins play the C major scale sixteen times, ascending and descending over one octave, but with a wide variation in dynamics and articulations. There are many more things to investigate than I have done so far, but I have already reached several STRIKING CONCLUSIONS:

1. Different VSL articulations, specifically "Short nts." on one hand and "Long nts." and "Perf. Legato" on the other, result in huge volume differences, even though they are all notated at the same overall dynamic level (e.g., "mf").
2. The eight basic levels of dynamics (ppp, pp, p, mp, mf, f, ff, fff) are not clearly distinguishable. The audible distinction between adjacent dynamic markings is inconsistent: ppp><pp is inaudible, pp><p><mp><mf><f is barely noticeable again and again, but the difference between f><ff is suddenly much too large, and between ff><fff, it may be even larger. From bars 1 to 32, the same four-bar phrase is played with the same articulations, allowing for direct comparison of the basic dynamic levels. When comparing individual bars, everything becomes very clear (compare, for example, bar 2 with 6, 10, 14, etc., and bar 3 with 7, 11, etc.).
3. Crescendos and diminuendos do not have a smooth progression (even though a linear dynamic curve is displayed in the "Synchron Player/Control" mode). In bars 53->58, I wrote crescendos and diminuendos over eight notes – from ppp to fff and back. It’s mainly the three loudest notes that make the big difference, leaving too little happening in the other five notes. Whether I use written dynamic markings or hairpins makes no difference. This phenomenon aligns with what happens with the eight dynamic steps from ppp to fff. (Perhaps redefining the values for the basic dynamics could fix this?) What I find completely bizarre is this: I noticed that the differences are controlled by CC1, "Vel.XF". The Velocity levels in Dorico’s Play Mode clearly show this as well. But where the massive audible differences occur, the Velocity levels show the smallest differences!
4. The different types of accents create too large a volume difference compared to the chosen base dynamic. Play bars 33 to 44, and this becomes very clear. Bars 37-38 are particularly shocking. Just listen, you’ll hear what I mean. In bars 45->48, I aimed to keep everything within the mf level – including the staccatos, accents, etc. – they should be clearly audible but not excessively louder than the "regular" notes. To make this sound acceptable, I had to lower the base dynamic of the "special" notes by no less than 3 to 4 levels! (from mf to pp or even to ppp).
5. There is hardly to no audible distinction between several articulation markings. The difference shouldn’t be massive, in my opinion, but it should be noticeable. After all, if this didn’t matter, there would be no reason to provide so many articulation marks. In bars 33 to 44, I applied different articulation markings to similar notes. It’s interesting to compare notes across different bars. For example, there is no audible distinction between the "Accent" and "Marcato" marks (compare bar 33 with 41). Similarly, placing a "Stressed" mark makes no difference to a regular note (compare the last two notes of bar 35 with those of bar 43). Comparing the same notes with those in bar 39, which are marked with "Tenuto", I hear a miniscule difference, which is, in my opinion, just a fraction too small to be meaningful. (For example, compare the B-A notes in bars 35 and 51.) And another issue: bars 36 and 44 feature similar notation, with and without "Tenuto", for the E-D notes. But something strange happens here. The next two notes are identically notated in both bars, C-G with a "slur". However, the "Tenuto" in bar 44 apparently provokes the following "slur" to be played significantly louder than the one in bar 36. I find this inexplicable.
6. In the Synchron Player/Control module, I see "Dim.Ctrl/B" as a third parameter, but it lacks a CC designation. Instead, I see "Speed". For most notes, this value hovers around 33, but for some notes (typically at the beginning of a phrase), the value is zero. I would like to control this parameter from the score because I usually want to start my phrases with more decisiveness. But I see no way to control this parameter from the score. Therefore, I have to use a rather extreme workaround to eliminate this unwanted effect, resulting in an awkward notation in the score.
7. Expression CC11 is not utilized at all, despite its potential to greatly enhance nuanced expression.
8. If I want a phrase to sound "normal" (as I would play it when looking at the score), I have to resort to a lot of "tricks" given the current situation. To illustrate this, I added bars 61 to 64 at the end. I tried to make the standard phrase (as played eight times in bars 1 to 32) sound as neutral as possible in terms of dynamics. I aimed for an expressive mf rendition, without being excessive. Regarding the "tricks": bar 61, first note – the goal was to achieve a "normal" start, i.e., to neutralize the "Speed zero". All subsequent dynamic and phrasing marks within this "normalized" passage aim to neutralize the inconsistencies in the Expression Map. Aside from these interventions, which should ideally be avoidable, there are some positive aspects worth mentioning. The way the staccato quarter notes in bar 61 contrast with the eighth notes and rests in bar 63 is very clear – it sounds exactly as it should, IMO. Similarly, the difference between the slur over 4 notes in bar 62 and the two slurs, each over 2 notes, in bar 64 is also spot on. Notably, in bars 45 and 46, the distinctions between all articulations are correct, although I had to notate ppp to eventually achieve a mf sound.

MY GENERAL CONCLUSION:

It is an unrealistic goal to make my music sound in Dorico as I intend it from the score with this Expression Map. To achieve a satisfactory sound, I have to do an unreasonable amount of extra work, which moreover ruins my score. Therefore, this is essentially a no-go. I suspect that I can adjust the Expression Maps to suit my needs, but I’m unsure if this would actually solve the problem. And I don’t yet know how to do that, either. If you have successfully modified your Expression Maps or adjusted settings in ways that have solved similar issues, please tell me. I hope this topic can encourage people to guide me in the right direction – and perhaps this can also prompt VSL to improve the Expression Maps. I know many consider Dorico to be a notation program, but I bought it as a composition program, where I can also create demos. If Dorico, in combination with VSL (which I purchased because of its high-quality sounds), cannot deliver this, and I need to use a second program to achieve it, this complicates the process enormously. I would find this very unfortunate, and it may lead me to explore other combinations of notation software and sound libraries that indeed might achieve this. Thank you in advance to everyone for your positive contribution.

[mducharme]

@DirkMM said:
Different VSL articulations, specifically "Short nts." on one hand and "Long nts." and "Perf. Legato" on the other, result in huge volume differences, even though they are all notated at the same overall dynamic level (e.g., "mf").

Most of these complaints are complaints about the library itself and not Dorico. I believe VSL said that they prefer to make the staccatos louder than the long notes in their programming because performers have a tendency to give staccatos a slight accent. But you can always go into the Synchron player and rebalance the individual articulations with respect to each other. (Although I don't usually do this myself - if I need them to balance with longs I will usually just open up the velocity lane in the MIDI editor and use the pencil tool to decrease them, takes a few seconds)

The crescendo in 53-58 sounds pretty good to me - I'm not sure what you're complaining about. You can see the curve that Dorico gives it at the bottom:

VSL is some of the most consistent in this way in terms of the programming, but there are still some minor inconsistencies here and there.

Dorico itself is not the problem - you can see it's doing a pretty standard curve there.

[DirkMM]

@mducharme said:

@DirkMM said:
Different VSL articulations, specifically "Short nts." on one hand and "Long nts." and "Perf. Legato" on the other, result in huge volume differences, even though they are all notated at the same overall dynamic level (e.g., "mf").

Most of these complaints are complaints about the library itself and not Dorico. I believe VSL said that they prefer to make the staccatos louder than the long notes in their programming because performers have a tendency to give staccatos a slight accent. But you can always go into the Synchron player and rebalance the individual articulations with respect to each other. (Although I don't usually do this myself - if I need them to balance with longs I will usually just open up the velocity lane in the MIDI editor and use the pencil tool to decrease them, takes a few seconds)

The crescendo in 53-58 sounds pretty good to me - I'm not sure what you're complaining about. You can see the curve that Dorico gives it at the bottom:

VSL is some of the most consistent in this way in terms of the programming, but there are still some minor inconsistencies here and there.

Dorico itself is not the problem - you can see it's doing a pretty standard curve there.

Thanks for your reply. I agree and also believe that Dorico is not the problem. Therefore I posted this topic in the VSL forum.
You said: "But you can always go into the Synchron player and rebalance the individual articulations with respect to each other. " I didn't know of this possibility, but I was hoping that something like that would be available. Can you tell me how to do that? And I hope it can be done for all cases, that is, that I don't have to repeat it for every single note. I hope, after performing the adjustments, that I can save them as my own "user presets". As a matter of fact, if this were to be possible, I guess I could also create a clear distinction between accent, marcato and stressed, all in proportion to a plain note. Right?
The curve that you show has indeed a normal evolution, so indeed, Dorico does his job well. But when I listen there undenyable irregularity - like I described, the 3 louder notes realize 2/3 of the crescendo. Do you really here a normal crescendo? If so, there must be another reason why my sound differs that much from yours...
And BTW, isn't it possible to make personal adjustments to the Expression Maps? Maybe that would be an even more powerfull method to achieve my goal, than making adjustments in the Synchron Player.

[mducharme]

@DirkMM said:
"But you can always go into the Synchron player and rebalance the individual articulations with respect to each other. " I didn't know of this possibility, but I was hoping that something like that would be available. Can you tell me how to do that?

You select something like "Short Notes", go to "Edit" and bring down the fader. This will make the short notes quieter. If this adjustment for short notes makes, say "Spiccato" too quiet, you could go down one level into the Spiccato folder, and bring up the fader there. So you have fader controls over multiple levels.

Unfortunately, the fader affects all dynamics equally, so in the event that the loudest notes are correctly matched and only the quiet notes are too loud, bringing down the fader may fix the quieter notes but make the loudest notes too quiet in the short forms. There will not be a good fix for this that I am aware of, because currently VSL does not seem to give access to the crossfade curves between layers or amplitudes for individual layers. I am hoping that we will eventually get more powerful curve editors similar to what is present in the Synchron Pianos.

@DirkMM said:
like I described, the 3 louder notes realize 2/3 of the crescendo. Do you really here a normal crescendo?

The crescendo sounds fine on my end as I said, but I'm also aware that Duality Strings Regular has no layers below about pp or p, so once it gets down to somewhere around pp or p it can't get any quieter really (CC1 values between 1 and about 30 or so will all sound pretty much the same), and so I know that and given those constraints it sounds pretty good. Since you are coming from Cubase, I would have expected you to be aware of these limitations, since you would have had them in a DAW as well. The official solution is either to use CC11 automation or timbre adjust automation to bring the volume down once it gets down below p, but this has to be drawn in manually in the CC curve editor in Dorico and cannot be automated at the present time. I had previously asked VSL for a "niente" feature like some other vendors have to handle these quieter bits, where it would automatically fade out to nothing at the bottom end, as a way to solve this bottom layer stuff, but they seemed reluctant to implement this type of feature.

If you have Duality Colors as well, there is an alternative way of handling the lower layers - you can use the flautando legato for p or lower and then switch to ord. above p, by adding playing techniques to the passage in Dorico and hiding them so they affect playback without appearing in the score. This allows you to get quieter dynamics than p without having to go into the CC editor lanes in Dorico.

@DirkMM said:
And BTW, isn't it possible to make personal adjustments to the Expression Maps? Maybe that would be an even more powerfull method to achieve my goal, than making adjustments in the Synchron Player.

This would be the case if expression maps were very powerful, but they are not. Expression maps are designed with the idea that vendors have already pre-balanced all of the articulations with respect to each other correctly within the player itself. If this is not the case they have very limited capability to rebalance articulations.

The only per-articulation control that you have in Dorico expression maps that has the capability to rebalance articulations with respect to each other is the range for the velocity values. For instance, here I could go into Staccato in the expression map and set the max to 100 instead of 127:

But in practice this would not be a good solution as it then prevents you from using the loudest dynamics of the instrument, so I do not use this myself, as tempting as it is.

If they allowed dynamic curve control per articulation, this could be a solution, but currently they only provide a single dynamic curve control for the instrument as a whole (all articulations together).

You could ask the Dorico team to provide this dynamic curve per articulation, but they would probably tell you that you should just contact the vendor and get them to balance their articulations properly in the sample library programming to begin with.

Expression Maps are not as powerful as you seem to think - they are actually quite simple. You might be expecting much more intelligence there than actually exists. Making them too powerful would reduce performance too much in Dorico, even for note entry operations, so they have to keep them simple and limited in scope for those reasons. The player is often the best place to implement these fixes and not the expression maps.

[mducharme]

I would also add to that - in many cases in full orchestral textures, a lot of these inconsistencies will end up not really being noticeable. You might be getting too much "into the weeds" at an early stage, trying to get the fine details right, but losing your sense of the big picture.

In practice it only takes a couple seconds, if staccatos are too loud, to run the pencil across the velocity lane in Dorico to decrease the velocities. It is manual, but it is so simple, it is almost not worth the struggle to go through all of this fine balancing.

Where your efforts should be more focused is on getting the entire orchestra balanced with relation to each other, and balancing the individual instruments/sections relative to each other. That's the important part, and it takes some time, and it takes having some pieces that you have good recordings of that you can load up and hit play to test the balance.

In my case my template file has 6 flows with 6 different orchestral pieces that have multiple good recordings so I can fine tune the balance. When I make adjustments I check with all 6 to see if it made things better or worse overall.