Hi Andy, Thanks for responding. Yes, I also think that could be usefull. But it's not the end of the story yet ... ;-). Up to now, I adjusted it all by ear. Now I'm also looking at the velocity layers to see if it also makes sense from that perspective.

In the whole Special Edition Plus library, there are just 6 different configurations for velocity layers. One being unique for the crotales only. Another being unique for the timpani. So, actualy, only 4 velocity configurations are to be in concideration, which are:

3 velocity configuration for strings, woodwinds and brass:

2 layers with p = 0-88, f = 89-127
3 layers with p= 0-55, mf = 56-108, f = 109-127
4 layers with p= 0-55, mp = 56-88, mf= 89-108, f = 109-127

1 velocity configuration for drums, cymbals and gongs:
6 layers with pp= 0-30, p=31-50, mp=51-70, mf=71-90, f=91-110, fff=11-127

In order to get optimal results, it would be very interesting to know a bit more about how velocity cross fading is implemented. To be more precise; 1) what's the bandwith for each velocity 'breakpoint' (sorry, a simpler word doesn't come up) and 2) is crossfading implemented lineair or ... ? Or in other words, 3) with which velocity values do I hear only one sample and with which values do I hear the cross faded samples. And 4) how much are they cross-faded? 5) Do all the instruments with the same velocity configuration, have the same cross fading implementation?

This is especially important to get the horns sound better because they are the only patches with 4 velocity layers. Which makes the velocity values unique in the range of mp and mf. For example, the default value in Sibelius for mf is 84. But the mf velocity layer for the horns is triggered at 89 and above. Which means that with the standard Sibelius values, the mf patch in the horns will never be trigger, only in crescendo/decrescendo situations from mf to f and vice versa.

Of course, I can also do this by ear but some facts would be great to know.

Really hope you can help on this.

Abração,

Wim

Wim,
What you're relating here explains a lot about some behaviours I couldn't explain with dynamics and Sibelius. Thank you so much for sharing! If you come up with some more conclusion re: dynamics and Sibelius, please let us know.
Best,
Denis

Hi Andi!

OK. Will do it by ear then.

Abraços,

Wim

Here the Sibelius values that I end up with (for the time being).

dynamic, default, new value, %
ppp, 20, 0, 0%
pp, 39, 10, 8%
p, 61, 36, 28%
mp, 71, 56, 44%
mf, 84, 74 , 58%
f, 98, 91, 72%
ff, 113, 109, 86%
fff, 127, 127, 100%

Sibelius allows only for one value per dynamic symbol which are applied to every instrument. So, these values being kind of a compromis between:
- the velocity tables in the VSL documentation
- the differences in velocity ranges as choosen by VSL for the various instrument(groups)
- the velocity ranges used by other libraries I own (Synthogy Ivory at this moment)

I additionaly adjusted to values by playing my own scores as well parts of "well known pieces" and using my experience to tell me how the score would sound in reality.

Some observations by now:
- ppp doesn't really work well. I prefer to compensate this with by seperately lowering the expression value (and reapplying the 'standard' value afterwards!).
- With dynamic mf and above, the overall sound of the orchestra gets more dense then what the score represents. Could it be that the Special Edition just doesn't have enough layers to support that level of quality? Or is it, because in reality, the sound of all instruments gets absorbed better by the environment. Are there common 'tricks' to solve this, besides EQing or using MIR... ?

By now, using the complete Special Edition, I'm really very happy with the quality of my MIDI-mockups that come right out of Sibelius. Without additional editing the quality is quite amazing (as agree my collegues whom are professional and academic level musicians too).

Any suggestions, additions or comments are welcome ofcourse.

Abraços,

Wim

Hi Oceanview,

For me personally, I think it's an ongoing process to get the best out of VSL. And my observations are quite similar as what you mention concerning the in-balance of the samples within the VSL standard matrixes. Staccato being an example, but pizzicato and col legno also, as others ... .

I think, partly this is a matter of taste, partly it's a matter of context (using the samples in a small setting or in a large ensemble), partly it's a matter of accoustics applied (EQ, reverb, etc), partly it's because the limitations in controlling the play back by music notation software, partly it's because VSL isn't an orchestra in a box ... .

This last observation (VSL not being "an orchestra in a box") is interesting I think. When I bought the Special Edition I thought I bought a "virtual orchestra", ready to be played. Later, I realized that I bought a large quantity of 'raw' sounds and had to build up the orchestra all by myself. This being a wonderful adventure, a great test for my ears and knowledge and absolutly valuable from a point of view of personal development. But, in the end very time consuming and from a composers perspective totaly in-productive ... . So, the strong point of VSL to my opinion is it's flexibility, but from a productivity point of view it's a uge weakness.

So, in the end I think it would be great if a small number of high quality templates would be available just to get good results right from the beginning (the templates by Beat Kauffman of CreaTech being an example in this direction). Such templates could be grouped by "DAW/notation software", "musical style" (romantic, contemporary) and by "ensemble type" (small, medium, large) and should include all the settings at Vienna Instrument level, at Vienna Ensemble level, and at "DAW/notation sofware" level. Addionally, some demo music should make clear the expected quality & limitations of that configuration.

Nevertheless, I'm very interested in some of your observations and what kind of values you use to balance the patches within a matrix. I think the idea of representing this graphically as a grid could work very well. So, let's give it a try?

Abraços,

Wim

Hi Andi,

My observation is that with the above table applied, only ppp doesn't really work well. Could this be solved by changing the value of Dynamic Range?

The manual states the folowing:
"With Dyn.R., you can adjust the dynamic range of a Patch, i.e., the difference in audio volume between MIDI velocity 0 and 127. The setting can have a value from 0 to 100, and marks the volume difference in dB (decibel) for the lowest velocity – if Dyn.R. is set to 16, this will result in a volume increase of 16 dB for MIDI velocity 0, while 127 stays the same.

If I understand this correctly, lowering the dynamic range would result in less volume when using low CC11 values (using velocity crossfading for example). I tried this for one patch but lowering the value seems to increase the volume, while higher values decrease the volume ... .

Because changing these values for all the patches, and all the matrixes would mean a lot of work, I prefer to know some more details first ... ;-) I also think changing these values could result in a complete inbalance of the patches within a matrix ... which is not really what I'm looking for ...

Abraços,

Wim

Olá Andi,

Thanks for your answer which are in line with the behaviour of VI.

BTW: A suggestion: to change the description in the manual a bit. I read some posts on this community which are a couple of years old about Dyn.R. which can be related to the misunderstanding of what is written in the documentation.

Thanks for your reply and still working on a good solution for "ppp" ... ;-)

Abraços,

Wim