Chris Hein dynamic layer problem

I recently purchased a few of the Chris Hein libraries and it wasn't until I played a bit on the cello library that I went back and realized I am having a strange problem on ALL of the libraries:

If you are familiar with the libraries you will know that they all have a Dynamic section where you select the dynamic layer by one of several methods. I have noticed that only the "Keyboard" selection actually gives you a normal sound. Setting this to X-fade, Keyb. & X-fade, or Auto X-fade gives the sound a kind of weird synthesized quality. This applies to every single articulation on every single library, basically making Keyboard the only usable dynamic layer selector.

I am really really hoping that this is a problem that has some remediably cause. Anyone have experience with these libraries and can tell me whether yours does the same thing? It's particularly noticeable on lower notes in whatever library you are using, but it's present throughout the entire range now that I know to listen for it in the upper registers.

 

I have contacted Chris and his answer to me wasn't very reassuring.

According to Chris, this is the phase alignment technique he uses. Basically it means that the sound on any of these x-fade settings will not be normal and will have this buzzy quality to them that is not entirely pleasant, especially on lower notes. He tells me I'm the very first user to complain (or apparently even notice) this, which kind of blows my mind.

Essentially, if anyone is considering buying any of the Hein libraries, consider whether you are 100% happy having the velocity of your notes be the only way you have of controlling which dynamic layer triggers. If you were expecting to use a mod wheel or some other expressive device like breath controller to shift between these layers as you play, the sound quality of the samples will suffer.

 

Your comments here don't reflect my usage of either the brass or woodwinds libs.

 

It's only because you haven't noticed it yet. As I said, Chris Hein himself told me this is the way it will sound.

If you haven't noticed it, do the following: put headphones on, select for example the flute in the woodwinds library. Have it set to keyboard, listen to that sound (especially on the lower notes, where it becomes glaringly obvious), and then switch it to x-fade. No offense, but if you can't hear the difference, you don't have very good ears.

 

Wow, Ill get back to my ignorance then.

 

That's your prerogative.

 

The idea with phase locking is that xfades will sound smooth and without the notorious doubling effect that can and will happen if the phase between different velocity samples isn't matched.

I would take no doubling effect any time of the day or night.


Also cut the offending comments.

 

I understand perfectly the idea behind the phase aligned samples. What I'm describing is what they actually sound like.

I'm really happy for you that you like them.

 

Wow, I noticed this very problem right away as I first tried the Chris Hein Solo Strings, and searched to find an answer. It's unacceptable! I play my strings using the mod wheel for dynamics and consider velocity mapped "long/legato" strings to be the "wrong way" of doing it. The buzzing sound is too annoying to just live with! Yet, velocity mapped strings can't have smooth transitions, and it sounds aweful trying to play live. Hopefully it can be fixed with an update.

 

Yes, the norm for string libraries these days is to control dynamics using the mod wheel. With the Hein libraries (all of them) this is simply not possible if one does not wish to use the "phase aligned" sound sets (basically all the settings except "keyboard" to select dynamic layer). I still find the library usable but basically it requires a much different way of approaching dynamics and many things you can do easily with almost every library under the sun are difficult to impossible to do with the Hein libraries.

I spoke to Chris himself about this and he contended that I was "the only one" to ever complain about this. That is, to be frank, either an outright fabrication or reflection on how many people use his libraries. I didn't want to press the issue with him and make an enemy, but I did post some comments reflecting my experience using the library in this regard on several videos about his libraries and ended up getting a letter from him expressing displeasure that I had done so (so I probably don't really have a friend in Chris Hein anymore, regardless).

He also has told me that there is basically nothing to be done about this. He showed zero interest in the prospect of correcting or changing this with any update, so unfortunately I would not hold out much hope for that. I also found another very fixable error in one of his trombone libraries which I contacted him about and even proposed a relatively easy way to fix it, suggesting if he did an update that I would include it. He tersely said he would look into it, after which I never heard back from him. It's clear to me that he is not interested in working on these libraries anymore and I would not wager on further updates.

I'm not trying to "smear" his company or himself here. Everything I've said is 100% factual and you are free to private message me and I can provide the full e-mail correspondence between Chris and I if you believe that I am distorting his words or the context in any way.

 

Bummer! I understand your frustration, I will contact him myself about this, maybe there is another work around. I know nothing about the Kontakt scripting behind the issue, but it may be just the way the library is designed and impossible to fix. I understand the idea behind the CH Libraries is programming to achieve complex lines, but I'm so used to using the mod wheel for dynamics, it's difficult to play and compose. I can't get consistent or accurate velocity playing on my midi keyboard, so playing in velocity mode has notes jumping from velocity 2 to 4 o 5, etc, very annoying. I've resorted to using the CC7 (Main Volume) to fade out individual notes, which is definitely the incorrect way to do it.

 

Unfortunately, the way I understand the library is that there is literally not way to "fix" it, which is to say that the phase aligned samples sound the way they do because that't the way they're made. There is no way to get the phase aligned samples to sound any better, I think. Now that being said, there are other things that could be tried:

1. As you say, one way to do this is to automate volume. Interestingly, you could potentially assign a control to the "volume" slider within the Hein library except as you'll quickly discover, the value for the volume is acquired at the time the note on message is received, which means that changing that volume parameter after you press a note has zero effect. Now, there are other ways of getting the volume to automate, either by using the actual volume slider for that Kontakt instrument (the one with the split diagonal white/grey triangle and the - and + at each end) or using the actual track volume within your DAW. As you say, however, those aren't really ideal for several reasons, not least of which is the fact that you aren't actually triggering different velocity layers which have differing timbre, you are just making one velocity layer louder or softer. Not very good as a solution.

2. The other potential solution (and this would require an update or at least someone good enough with scripting and the ability to alter the Hein interface) would be to allow for selecting original sample sets vs. phase aligned sample sets for the other methods of dynamic layer selection (i.e. x-fade, etc). Now, according to Chris, the reason this isn't already how the program works is because you will hear phase anomalies as you sweep through the dynamic layers because the phases might not line up. I'm partially dubious of this, even though in principle I understand it. My doubts stem from the fact that I think this is how most string libraries do it and I have not been unsatisfied with the way they sound, if they are in fact producing small phase inconsistencies. In other words, how does Spitfire of LASS accomplish this same feat without the use of phase-aligned samples (which they certainly don't)?

One test you could do, which is a bit elaborate (and now I'm tempted to try) is to set up multiple Kontakt instances of the same Hein instrument and send them the same MIDI notes simultaneously but at different velocities (specifically velocities designed to trigger different velocity layers of the samples - for this you'd have to look in the mapping panel and see where the break points are). Then you would have to find a way of cross-fading in your mix between each of these Kontakt instances as these notes sound simultaneously. Can you get a pleasing sweep from the lowest velocity layer to the top one using only volume crossfades? If so, then this should be implementable as a version of number 2 solution above. At the very least, it should not be difficult to implement as an update: pretty much he'd have to add a switch to the dynamic's section that essentially turns phase-aligned samples off and uses the "real" samples for all dynamic layer selection methods. Whether it would sound good, I don't know yet, but the above test would reveal this.

If you talk to Chris, feel free to drop this knowledge on him. Maybe you will be more persuasive than I managed to be. I think he doesn't want to hear from me anymore, so it's probably up to others to convince him that this is a real problem.

EDIT: I just performed the above-mentioned test myself with interesting results. First, my methods were pretty quick and my ability to crossfade between layers in this experiment was pretty limited and basically entailed me trying to draw even volume curves from -inf to 0 and then back again to -inf as each velocity level triggered on a separate instrument playing the normal samples (i.e. not phase aligned). The curves I drew were not great, and so you hear some swelling and moments of lesser volume as each layer transitions (i.e. the volume doesn't stay constant as it would if this were implemented better). The real purpose of the test, though, was to see how these non-phase aligned velocity layers would sound transitioning from one another.

The results are, to me, unequivocal: it sounds FAR better having a normal mixed transition between these normal samples than it does to use the X-fade on the phase aligned ones. The experiment DID expose a few problems with the samples in the lowest velocity layer (the "modern cello" has some weird clicks which sound like metronome count-ints that occur early in the sample... these would have to be corrected, ignored, or worked around in some way - I am not at a point of proposing a real solution to this yet. One thing at a time).

I think it's really worth either trying to contact Chris to see if he will at least grace us with the ability to choose whether we want to use the phase-aligned samples even if he is convinced they are the best, or to use the normal samples regardless of dynamic layer selection means. If he will not comply, I do think that this is something that can be accomplished without his help but it would require probably someone who knows more about Kontakt than I do. It may be as simple as finding the sample sets that are phase aligned and deleting them, replacing them with a copy of the "real" sets but this means you're using twice as much RAM as you would have to if you had no intention of using the phase-aligned sets. The best solution would be to actually change the way the library interface behaves in regard to it auto-selecting the phase-aligned sets when using the non-keybard velocity selection methods. Then you could purge the phase-aligned sets and have less RAM footprint and not have to deal with those awful sounding phase-aligned samples again.

 

I noticed the exact same issues as you stratoculture666 with X-Fade. It's pretty much a useless feature of the solo string instruments. It sounds terrible. To me it sounds like this initial attack samples are not triggering...not to mention the terrible sound of the buzzing. I just don't buy people haven't noticed this...more likely just haven't let Chris know.

If you haven't, download the Cubase demo session from Chris's site and you will notice that almost all of the session DO NOT use x-fade. They automate the volume of Kontakt player via CC 7 to achieve dynamics. While this work's around the issue, it doesn't always sound right because the long note phrasing has natural swelling in it.

I also have experimented with fading the samples myself and while it's not perfect, it sounds worlds better than the X-Fade feature. This is the approach I am going to take when needed.

The instruments out of the box sound amazing and all-in-all I am very happy with the purchase. I still think its one of the best solo string libraries out. However, it completely baffles me the vast quality difference between the Keyboard velocity and X-fade mode sound. If I were Chris, I would have just left out X-fade altogether.

I'll reach out to Chris as well, but I don't have much faith that X-fade will be fixed or improved.

 

Thanks Rick, I appreciate others coming forward and supporting the fact that I'm not imagining this

Yes, one thing that Chris did admit to me is that pretty much all the promo material and demos for the libraries use "keyboard" as the dynamic layer selector. I don't know how else to interpret this than he must himself realize that X-fade doesn't sound as good, even though he feigned surprise that anyone would complain about it.

I agree that the libraries are overall quite good and they continue to be my go-to choice for solo strings, woodwinds and brass. All libraries have their limitations but I find his products to be quite playable and the natural dynamics of each sample (particularly the long and short expressive patches) make some automation that you would ordinarily have to do less necessary to get them to sound musical. I do still wish you could do swells more convincingly using the mod wheel like practically every other library under the sun.

I have been gradually going through all the libraries and swapping the phase aligned sample maps out for the normal sample set in the appropriate groups so that when "x-fade" is selected you are hearing the normal samples instead. It's not a difficult procedure, it's just a bit tedious. The end result is actually not bad. I don't hear phasing issues in most cases anyways. Maybe a better ear than mine would pick up on it, but I find myself not bothered by it at all. I still mostly use the keyboard dynamic layer selection most of the time but when I have a line that calls for swells I will now be able to do it.

I do support people reaching out to Chris. Prepare for him being a bit defensive about this and also not interested in changing anything, but as I told him, it's important for people to be able to provide feedback about a product they spent money on.

 

That's a great idea about remapping the samples. I've not done this before so...I'll do a bit of research on how to do it and experiment myself. Do you know if you can export the mappings? The reason I ask, is that I wouldn't mind providing it as a download for those people like ourselves that are looking for a better sounding x-fade option.

 

Re: exporting the mappings, basically you would just save whatever patch you're working on as a new .nki file. I don't know if you can save a mapping itself but even if you could there are so many within each patch that it would be cumbersome to the point of impractical to share those individually.

Re: how to do this, I'll try to give a brief explanation:

1. Load a patch, let's say for example CH - Solo Viola.nki from your library menu at the left
2. Click the wrench icon
3. Select both Group Editor and Mapping Editor

What you are looking for are any groups with XF in the name, for instance in the example patch the first one is Blend 1. The normal samples are Blend 1 VK and the phase aligned samples are Blend 1 XF P1, Blend 1 XF P2, Blend 1 XF P3, Blend 1 XF P4, etc (there is also P5 and P6 but these are blank and you don't need to worry about them).

4. Click on the VK version of whatever group you are working on and look in the map editor. Select everything and copy it.
5. Click on the associated P1 group and delete everything.
6. Paste
7. Delete everything except the bottom velocity layer
8. Select all of this and drag the upper boundary of it to the top. What you have done here is substitute the lowest velocity layer of normal samples for all velocity layers of the X-fade (meaning, if you have the X-fade controller on layer 1, it doesn't matter what velocity you are hitting on your keyboard, it is going to still trigger layer 1)

Now for the other layers

9. Click on the associated P2 group and delete everything
10. Paste
11. Now delete everything except the 2nd velocity layer
12. Select all this and drag the lower boundary to the bottom and the upper boundary to the top. You've now done the same thing as above, where now, if you have the X-fade selecting velocity layer 2, you will be playing layer 2 of the samples no matter what velocity you are playing on your keyboard.

You will follow the same procedure for layers 3 and 4. Note the following things:

- Most groups will have 4 velocity layers and therefore 4 XF groups that you will need to replace. Some may have different numbers, but I don't think you ever need to be concerned with more or less XF groups than there are velocity layers you will have from the VK group you are copying from, if this makes sense.

- Sometimes you will encounter an XF group that doesn't have phase aligned samples in it. You can just ignore these and leave them unchanged, as they are probably already set up exactly as you would if you followed the above procedure. You will know that the XF group is phase aligned because in the mapping editor, you'll see the sample names in the map ending in "_p". Actually the above example of "Blend 1" is a bad one because this is one where you don't need to do anything because the XF layers are not phase aligned. The first one that actually needs your attention in Solo Viola is 01 Sustain VK (if you look at 01 Sustain XF P1 you will see that the lowest sample is VCA_Sustain_c2-1_p), indicating that this group is phase aligned samples and you should replace that map using the procedure above.

- Any of the "short" sample groups, i.e. staccatos I would recommend not changing, even if there are phase aligned samples. These are patches that, in most other libraries, would also be triggered using keyboard velocity and not the mod wheel as most sustains and legatos are. This makes sense from a playability standpoint because you will usually want your staccatos and spiccatos to respond to how hard you hit the key and you are not going to be swelling them anyways because that's not in any way realistic (I doubt there are real players creating volume fades mid-staccato).

- I recommend from the beginning just saving an alternate version of the patch with a new name (I call mine the same as the original with "NO PAS" at the end, indicating "no phase aligned samples") and saving periodically along the way. That way if you screw up somewhere along the way you won't have to start over and you won't be changing the original patch, just in case you still want that for some reason.

If you have any questions about this, let me know. At this point I have done both cellos and both violins. It will take me a while to do, but eventually I plan on doing the viola, contrabass and ideally all the brass and woodwinds as well.

 

Thank you...these instructions are awesome. I was attempting to just try one note all the way up the velocity layer, but this process makes it much easier. I'm going to give it a go here shortly and report back my experience.

 

So I just did this with the long dynamic articulation and it sounds SOOOO much better than the X-Fade version. I do notice some chorusing taking place as you move through the velocity layers due to the way the x-fade script is fading from one velocity layer to another. However, the sound is actually usable unlike X-fade with the phase aligned samples. The chorusing could be limited if I could change to a shorter cross-fade across each x-fade velocity layer boundary.

 

So I see that all 6 samples play at the same time with the volumes set to 0 for all the layers except the two where cc 11's value (x-fade) is located. It looks like there is cross fade script lowering the volume on one while raising it on the other. The problem with this approach is that you really don't want two samples playing at the same time the entire range of one velocity layer...that's what causes the chorusing effect. With the scripts page being password protected (understandable) there really isn't a way that I am aware of to change it to just cross fading at the boundaries of each layer. It's like we'd need an override of the scripted volume parameters.

 

Great! Glad you had success with this.

Yeah the thing is unfortunately that the cross-fade is necessary or you'd have an audible jump (likely even an actual pop) when you transitioned from one sample to the next. You could, as you say, probably change how fast this occurs but I'm not sure it would sound very good. If you end up obtaining a password to unlock it and want to try and report back your results I'd definitely be interested in hearing them!

As you say, this method isn't perfect and on some articulations it can sound a little weird but especially on the sustained patches I agree that it is at least usable. At the very least, it's another tool in the toolbox, so to speak I also find that in a mix with other orchestration the phasing isn't that noticeable and I can usually live with it. This has made me wonder how other libraries accomplish this same feat, because they are obviously capable of doing volume fades without phase issues. The simple explanation would be if there is only only one dynamic layer to begin with, I suppose. The other thing I could think of is if the volume fade only affects the layer that was originally triggered and another layer doesn't begin playing until it is triggered, unlike here where we have all layers being triggered and a fade between them as they are already active.