I don't work at VSL. But I can imagine that the problem is when you have to process 1000s of samples and that, starting with the recording and then the editing, certain differences arise between individual tones, as you are now discovering. I can only recommend that you try out other libraries. You will notice significantly larger differences - also in the sound between the tones. In other words, the regularity that you obviously expect is obviously extremely difficult to achieve. Just imagine if samples are recorded over several days. Maybe you noticed that an octave had to be recorded again... Getting the exact same position, the distance to the microphone is incredibly difficult. Even if the volume of the sound is correct, the sound may still be slightly different... Maybe there will be other objects in the room later... and that's just on the recording side...
------------------------------------------------
How do you as a user deal with this?
A) It seems that you are a user who most likely produces music via a music notation program. Even though I've been producing music (mainly with VSL samples) for 20 years now, such differences have never bothered me because I adapt each subsequent note to the previous one anyway. This can actually only be done perfectly in a DAW (and Dorico).
I recommend that, if you haven't already done so and the musical result is important to you, you export a midi file from the notation program and then produce the music in a DAW. If you want musical results, no note should be played as loudly as the previous one. In this respect, such minimal differences between the samples are not a problem.
B) If you want to work with samples, the basic requirement is that you have a relatively large tolerance. Because the samples are basically dead materials, if you, for example, play a new "a1 staccato" 3 tones later, the same sample is called up as before, and you also have, for example, exactly one legato (no warm, no sad or emotional one) and, above all you are completely limited in the number of ways you can play. In addition, you have all the little technical errors that you notice again and again.
So you have to accept all these limitations and mistakes and still try to play this completely limited instrument (samples) with such virtuosity that it becomes music. And I don't mean music with 0.25dB more of something, no - i mean just music.
The recipe is to keep the whole thing constantly moving: the tempo, the volumes, the articulations, the dynamics, the filter curves - just as much as possible. When you hear this, you probably realize that minimal differences within an articulation aren't that bad. Of course, it's bad when a certain sample has an error that you then recognize over and over again. Of course that doesn't work. VSL is really perfect in this respect. But this also means that the individual samples have to sound fairly neutral.
Of course, all of these constant changes should be done in such a way that the music ultimately comes as close as possible to what it could sound like in real life. All the users have to decide how far they want to go with the effort.
C) Personally, I push the effort to the point where about 90% of what is possible is achieved. For the remaining 10% you would probably have to put in the same effort as before. But what I do and what is definitely a gain in quality: I convert all the instruments into audio files (without effects) and then create the orchestral mix with these audio files. In this way, I can, for example, bring individual tones that are too quiet to the necessary volume - using mixing automation. Of course there are a lot more possibilities.
----------------------------------
I don't know if you like all of my answers. I simply told you about my access to the samples. I am thrilled without any ifs and buts and have been for over 20 years. If there are mistakes and inappropriate sound errors, I try to hide them somehow, or solve a section of music a little differently.
Some examples
All the best and a lot of success
Beat
