Hello. I just upgraded to Cubase 15. I have tried using the attack compensation feature in the expression map with Cinematic Studio Strings. As an example, I created a violin line with the expressive legato and medium velocity. The delay for this is 250 ms. I entered 250 in the attack compensation and my violin playback timing is way off. I had to cut timing in half to get the playback close to in time. Am I doing something wrong or is this feature just broken? Thanks.
1 Like
The expressive legato isn’t consistent delay in CSS - it depends on the velocity of the notes unfortunately. Certain velocity ranges have a certain delay - lower velocities are longer and higher velocities are shorter I believe.
Thanks for your reply. I deliberately set the velocity of the notes to the same value to make sure the delay is consistent. I literally have to cut the delay almost in half to get it in time. I have also tried different libraries and I am experencing the same problem. If I set the global delay in the inspector, it works fine. Obviously that is not a solution for an instrument per track. Thanks again.
Sounds like next advance improvement is supporting variable attack compensation based on another value like velocity (either linear or a set curve)
2 Likes
I can confirm the problem.
I was looking online to see if anyone else had a similar experience and, surprisingly, this is the only thread or mention that came up.
I have tested this feature with multiple libraries and articulations and it seems that the attack compensation value entered is being doubled internally by Cubase 15.
This seems consistent with longer articulations (like the slow legato in CSS having a delay of about 333ms, but the attack compensation that makes it feel tight is roughly 165ms) and with shorter articulations (like staccatos in CSS having a delay of 60ms but the attack compensation that makes it feel tight is roughly 30ms).
This is not a library-specific problem, I have tested this with some Spitfire Audio products as well.
This is also not an error on my end either since I know these libraries extremely well and what to expect from them.
I repeated the same tests with global track delays as well as within Dorico and both of these tests worked fine with the expected values.
1 Like
This should be really simple to prove by using an instrument with a very fast attack time.
Here’s how I would test it.
- Create an Instrument Track using Retrologue with a simple waveform and minimal attack time.
- Setup an expression map for that track with an articulation using 500ms Attack Compensation (or something).
- Create a MIDI Part with a note event and apply the articulation created.
- Render the track to audio and use a Ruler Track set to Seconds to gauge the distance between the start of the audio and MIDI Event.
If all works as intended, you should see the Audio Event 500ms ahead of the MIDI note event.
All very good ideas. I think that Pandraos is right that Cubase is doubling the attack compensation time entered. I have now verified that this is happening on multiple libraries. Steinberg, please look at this. Thanks all!
Hello. In a previous post I documented that the attack compensation setting in the expression maps is not working correctly (delay timing is off). I have tried numerous libraries and the problem is consistent. I have to cut in half the required delay that the library needs to have the midi part play in time. As an example: I used Cinematic Studio Strings Expressive Legato for the violin 1 patch and fixed the midi velocity so I would get a fixed delay of 333m ms. I entered 333 ms into the attack compensation box, and on playback, the timing was way off. I cut the attack compensation in half (approximately 166 ms) and the timing was right on. I then rendered the midi track to audio. Interestingly in the audio track the first note played in time (even though legato was set even on the first note) and subsequent notes were over 600 ms late instead of 333 ms. So, the attack compensation appears to be broken. How could this not have been tested? I can live with cutting the attack compensation values in half and still come out ahead, but what a hack. Anyone else done further investigation into this? Thanks.
Confirmed, I’m seeing different results with CSS when I give the entire track a -333ms delay vs when the attack compensation is at 333ms. With the attack compensation the notes arrive too early. And yes with attack compensation set to 166ms I get about the same result as if the track had -333 ms delay so it seems like the attack compensation ms might be getting doubled.
I took the liberty to merge both threads, both talk about the same problem (unless I missed something?)
Thanks for the report, and apologies for the inconvenience. We’ll investigate.
2 Likes
I’m glad it wasn’t me doing something wrong. 
It looks quite obvious that the testing was not done on libraries that have any kind of significant delay. It is quite glaring. Thanks.
Thank you. This is something we orchestral composers have been waiting for a long time. And it appears to be broken… Disappointing… Thanks and please fix this as soon as you can so we don’t have to enter attack compensation values that are half of what they should be.
1 Like
With regards to legato…its often misunderstood that most sample libraries when engaged in “legato” mode via key switch, the instrument is performing different kinds of attack transients, typically note groups that have overlapping blocks on the piano roll are treated as inside a slur and the ones without this overlap are treated without legato transition, that would be the first note of a slurred group…or any other notes outside a slur when comparing to notation). The non-overlapped ones will have a much shorter attack compensation requirement then the ones after overlap. In the above you said 250ms which must be the expected legato transition for secondary notes, but make sure to realize that any notes that are on their own without overlap in front of them will not have that long attack time and so thus they would sound like they are way overly compensated. This will depend on your track and how the notes are grouped in legato phrasing with overlapping and non-overlapping. CSS goes even further and introduces different amounts of latency for the overlapped -slurred notes depending on velocity.
Currently Cubase really only provides a single attack compensation which IMHO is not all that is needed to handle legato modes of operation per the above explanation.
1 Like
Yes, I understand this. I have been using CSS for 9 years. The violin line I created was all overlapping notes with the same velocity, and the articulation was set to legato expressive. All velocities were below 64 which will trigger the slow legato (333 ms). Of course the 1st note is not legato, but the succeeding overlapped notes will be legato. With attack compensation of 333 ms, the timing on the line is way off. If I cut the timing in half, the legato line is in time, implying that Cubase is doubling the delay. If I set the global track delay in the track inspector, the violin line plays in time. The issue is with an incorrect implementation of the attack compensation feature. Also, two other people in this thread have confirmed the same problem.
Thanks for clarifying, I’m sure Steinberg will get the fixed soon.
Please consider adding an intelligent legato mode with different delay times for the first un-slurred note of a phrase and subsequent slurred notes because their delay times can vary substantially (over 200 ms).
Someone called Alex John Ernest Vincent wrote a Kontakt script for the Cinematic Studio Series with this feature which you could look into.
1 Like
Can you double-check and confirm that you’re not using any additional stage of time compensation in the project, such as MIDI inserts?
No. If I set the global delay in the track inspector, the timing is correct. Attack compensation, no. Multiple people have verified that Cubase is doubling the value. Attack compensation is broken.
In my setup, attack compensation appears to work correctly with both long and short-note articulations. Are you using CCs or key switches in your Output Mapping? Does the issue occur with both methods?
What I’m noticing is that attack compensation behaves differently depending on whether the notes overlap. For example, the sustain articulation works correctly when the notes do overlap, but falls out of time when they don’t. In the latter case, it seems the attack time value is effectively doubled (as already reported in this thread).
1 Like