JWL, thanks so much for this post. I have more trouble with legato violin section lines in VSL than anything else. Sometimes they work, other times I end up sounding like I'm using a 100.00 string library. I know it must be my lack of knowledge in how to use them, this helps.

Tom

@JWL said:

Drillan--

Have you played around with attacks and releases on the right side of the PERFORM panel? You might also want to toy around with what adjuments might work for you in the cell edit's ADSR window. You really have to chase these controllers around from one phrase to the next-- or even from note to note because what works in one place won't always work elsewhere. There really is no set-it-and-forget-it way of doing it, even though some of the Speed patches come closest to this in concept.

For as "compact" as VI has been designed, it really is not a turnkey setup at all. A legato patch or matrix "is" legato-- but "how" legato it might be for your purposes requires a considerable balancing act with x-fade and ADSR's in various combinations. Otherwise, notes will jump out at you at one moment or not be loud enough in the next.

For example, sometimes I want a mp or mf with vibrato, but the most active vibrato of violin layers is also the loudest. In these cases, I know that when the x-fader goes up, the expr faders must come down for the proper note volume.

I've elongated some of my release times on non-legato patches. Adjusting the attacks cannot be overlooked either, because a softer attack in the middle of a phrase contributes to the smoothness of the phrase. Also, I find the Universal Patches set up quite well for certain legato applications where I'd failed to accomplish the same effect with Performance Intervals. Keyswitching certain notes or series of notes between Universal and legato x-fade layers can help greatly. One must experiment with the x-fade at all of its levels before determining that the matrix or patch dosn't work. Again, ADSR settings play a big role with this when used in the right proportion. It's time consuming, but the results may be more gratifying.

Keep in mind, a lot of it depends upon exactly what you are doing. I was trying to do Barber's Adagio, and quickly realized just how many different layers of legato such a project required. I discovered by listening to several recordings that the violins aren't *always* playing legato. There is space between some notes even in a slow movement where the rest of the strings supported the harmony. The illusion was that the violin line was always connected seemlessly, but it really wasn't. In faster movements, VI shines because the user has access to a greater palatte of articulations which don't linger long enough on one type of articulation to generate ear fatigue.

Here's an exercise I did which might help you...

1. Set up a Vi Console with a wide variety of violin legato, Speed, Universal matrices and patches just to have all of your choices in one place.

2. Play in a violin line on a MIDI channel using just one patch without worrying about how it sounds. You may also want to add some accompanying harmonies with other instruments (even a piano) just to hear how the line plays against the accompaniment.

3. Create another MIDI track just for keyswitches. This will help things from getting too confusing, especially if you are use a DAW's notation features at any time. Isolating the key switches makes for easier editing, too. You can change the clef of the keyswitch track to minimize leger lines for extreme keyswitch octaves.

4. Try assigning your violin line to each of the patches/matrices and take note of which ones serve your violin line the best. Be sure to play around with x-fade and ADSR settings and take note of the benefits of these combinations for your current project or even for the purpose of using in them future projects. It helps to build a user "vocabulary", this exercise always leads to exciting discoveries that WILL come in handy sooner or later.

5. Begin assigning matrices and keyswitches in your keyswitch MIDI track, adjusting as needed. You may also want to record CC data on the keyswitch track or even create a separate MIDI track for this if it keeps helps you keep things clearer in your head. Sometimes one single track with all note and controller data gets confusing.

6. Save whatever matrices you've edited that serve your violin part the best, and then go back to your actual project. You may load those same edited matrices into your project with the confidence that they'll work as expected. It takes a bit of experimentation, but it's something that only has to be done once, until such time the challenge of finding the right combo of articulations arises again.

7. Last word of advice-- with something like VI, it is easier to find yourself listening to your mixes "with your eyes" and not your ears. Watching CC numbers and keeping track of nobs and faders can impair your hearing or psych you out. If you want to really narrow down what VI will (or will not) do for you, work with just 4-5 notes in the violin part that contain both small and large intervals both up and down. Assign each note or pair of notes to different combinations of patches or matrices to see just what effect they have and to determine which interval transitions you like best. Sometimes a great legato can be created without a legato patch-- depending on the result you want.

Good luck.

@Another User said:

JWL, thanks so much for this post. I have more trouble with legato violin section lines in VSL than anything else. Sometimes they work, other times I end up sounding like I'm using a 100.00 string library. I know it must be my lack of knowledge in how to use them, this helps.

Tom

Tom-- I only hope that my little homemade exercises do help in some way. I wish I could boast better results with the Barber because I really don't feel as though I've gotten my chops around the violin legatos by a long shot-- but in the midst of that "$100" sound, there were glimmers of promise with certain phrases. The thing that bugged me was that I'd spend about 3-4 hours twiddling with controllers in every possible mathematical combo-- then swapping out patches and matrices just to get 3 notes to work right. Of course, after six measures I and the day were both spent. The thing that encouraged me to start over with a fuller orchestral exercise was the notion that the strings, violins in particular, would not be nearly as exposed, and a variety of shorter articulations could also be employed.

But I'd really like to nail the Barber. In fact, I'm working from a Bernstein recording with the NY Phil-- one of THE slowest ever recorded, lasting at least 2 minutes longer than most other recordings. So far, I've found the need for about 26 different legatos across the entire string section-- unique patches for each rhythmic note value at each dynamic level (plus crescendos and dimuendos that behave the same way within the various legatos), as well as different amounts of vibrato or non-vibrato within each dynamic layer.

Though it's truly a geek-tweakfest, I'm not convinced yet that it's not worth the effort.

But hey-- if you make any sort of discoveries with ways and means to create any types of legatos of any sort, please post your finds. Having the Library of samples and factory patches is one thing. But it's the users' side library of tweaks that make all the difference.

Cheers, gents!!
J

@SyQuEsT said:

Did you try to use the preset files from the tutorials ?

SyQuest, I wasn't sure if your question was directed at me or to Josh (Drillan).

I personally always start with a factory setting whenever possible and tweak as needed. When the factory patch lacks what I hope to hear, I creat a custom matrix. But before I begin recording, most of the time with unfamiliar factory matrices, I'll play a note or two and move all the nobs around just to understand what the patch or matrix will do.

It seems unthinkable, but I'll even throw in an sfz patch for one note in a fast scale and turn it down just for one note. It doesn't sound like an accent, but it does sound different. I've done the same with a decresc. patch as well just for one note. You don't really 'hear' the decresc. per se, but when looking for variety, leave no stone unturned.

"I've just been listening to a LOT of classical CDs and film soundtracks just to re-orient myself away from computer behaviors and more into the real world."

Have you ever been listening to a score from a real orchestra and been struck by how fake it sounds? There have been moments when I hear something and reflexively think, "That's not working. He should have used a different patch." And it's a real orchestra!

I have a recording of Beethoven's Fifth by Toscanini, and the first time I heard it, I was convinced it was sampled because I'd never heard such quantized playing. But that was Toscanini.

You are right that all the lower instruments are slower to speak. And yet, the VSL flute, clarinet, and oboe are alto to soprano, and they seem more fliud than the violins. I've also noticed that my ear can tolerate the same connected sound in the woodwinds when the same intervals are repeated, but when I hear legato violins playing the same connectedness, it grows unconvincing more quickly.

We'll see what VSL does next.

You make yet another good observation, Plowman. So many recordings were just poorly engineered. In effect, it's all one form of "sampled" sounds, when you consider that live instruments are being reproduced and played back in electronic media.

Yes, Toscanini was a stickler for precision, almost to a fault-- but that was at a time when orchestral playing was quite sloppy in general. Tosacanini was the steward of the modern orchestral age, being one of the earliest conductors to make use of recording media.

For us these days, it's the "new" old argument that "perfection is not perfect". I can recall back in the late 70's an engineer explaining the wonders of quantization! Boy, was he excited at the notion that all human error could be removed. Okay, so we've survived the 80's era where electronic music was indeed robotic, at least in pop world. It had it's charm-- for about 30 seconds.

First we hear that "digital is going to revolutionize the way we experience audio". True enough, but ambition is always way ahead of reality. Suddenly, we're hear "analog is better than digital" for legit reasons. Just when we grew accustomed to this mantra, we heard "digital has gotten better because converters have gotten better". Now, it's more to the issue of certain elements of digital and analog must be combined for the best overall result. I'll buy that until the next mantra kicks in... (???)

Enter the performances of Toscanini. At the time, recording orchestras was still quite new. I think many subtleties were squeezed out of those early recordings, and I suspect that some of Neville Mariner's similar attention to precision would have not translated well had they been recorded at that time.

But then there was Leopold Stokowsky, who represented a whole different school of conductors in his day. He was a master at "phrase liquifaction" whose expressive qualities translated much better to early analog recordings than Toscanini's technical precision. What Arturo accomplished is not to be underrated by any means, but it's SO funny when the ear proves itself to be less wise than it thinks it is!!

Mistakes and ensemble goofs may not be desired with a real orchestra, but those elements certainly do nothing to convince a listener that the orchestra is anything but real. For me, it appears to be the risk factor of making an exciting musical phrase that *may* result in these errors. For many, the errors are quite forgivable if a certain "agony and extasy" is present in the performance.

So, what's a VSL musician to do? We have the ability to nail a Toscanini-type realization, but a large part of using VSL is revealing itself in areas that deal with what NOT to do as much as anything else-- don't quantize, for example. Funny to think that to make VSL sound "better" one has to make it sound "worse", if you take my meaning.

Very well said, JWL.

"Boy, was he excited at the notion that all human error could be removed." You have to love this. Forty years later, we're asking Herb for chair squeaks and room noise.

"We hear 'analog is better than digital' for legit reasons." Yes. I put most of my strings -- VSL and Garritan -- through a match EQ taken from the original recording of "Gigi." It's a CD, but the master was analog, of course.

"I'll buy that until the next mantra kicks in..." Perhaps, "It'll never be the same as hearing it live." Let's hope that's always the truth.

At the end of your post, you allude to the pedagogic value of samples. What's often lost in discussions on virtual orchestration is its capacity to focus the user on every nuance of the medium.

"every nuance"--- Indeedee-weedy!

So much has been taken for granted with earlier orchestra libraries especially. Even before you started using them, there was an unspoken understanding that there would always be a lack of humaness in the result, so it was not something that had to be consideed.

With VSL, a composer/orchestrator cannot avoid reconciling his conducting skills with his writing skills in conjunction with reconciling his computer geek skills to make it all work. Even with those elements reaching a level of equity, it becomes important to actually notate string bowings and articulations even for scores that will not be played "live". For me, anyway, it helps keep the required articulations in order and at a glance.

One challenge is distinguishing 4 sixteeths on one bow from 2 or more bows. For example, the first 5 notes of Rachmaninoff's "Rhapsody on a Theme of Paganini" have a particular sound. In most performances we hear 2 bows (4 notes + 1). The attacks on all five notes are quite different, but how it is bowed can be easily heard. Understand how it is bowed forces one to consider which samples are used and how they are used-- if the intent is to produce a believable result.

I suppose all of my latest posts are the result of an "uneveness" among the many skills required to use a sample library like VI/VSL. Using a score like "Salome" takes some of the pressure off, but the uncharted waters of VI still feel rather deep and trecherous!

Yes, life was simpler with Vitous and the K2500. They asked so little of us. Once we only had blue. Now we have cyan, teal, aquamarine, sky, royal, and turqoise. No wonder the artist in us is called to the fore.

"Using a score like 'Salome' takes some of the pressure off...." I'm not sure what this means. I assume you mean Strauss' work. Does the grandeur and size relieve you of the some of the detail work?

As I continue in this field, I am faced with a growing, reluctant admission of what computer orchestras can never be.

Thank you for all of your advice. I did what JWL suggested and got a bunch of great ideas.

I've found that the perf-legato_sul patches sound good under most circumstances when the violin is the lower portion of its range. When it hits the upper register, however, it sounds a bit closer to portamento than I'd like, so I generally switch to the standard legato patches for that (with a slightly lowered attack length), though it still isn't as convincing as I'd like. It's way better than what I had, however.

And on the subject of precision, it's the same as it is with computer graphics. The less precise you can make a rendered object look, the more accurate it seems. Adding a little bit of noise to simulate film grain can make all of the difference. It's definitely true that the eye (and the human ear) is trained to pick up on imperfections. Things without them seem unnatural...and it's usually because they are.

I'm still having difficulty getting good dynamics wiht my string lines. Moving controllers around for velocity crossfading just seems so...I'm going to use the word "unnatural" again. I'm not good at switching to the cresc/dec lines yet, so there's definitely some room for improvement there.

I am, however, still very happy with my purchase of Strings I. It's convincing to me at best, convincing to many at worst (at worst after some effort), and miles above the crappy string sounds in my QS8.

Thanks for the (continuing) advice and discussion [:)]


Josh

Hey Josh--

I'm so glad that any of my VI madness has been a help to you at all. Often, I feel quite alone with learning these interfaces. Many have rightfully asked for FXP or AUPRESET and MIDI files to see *what* was done and *how* it was done, but I'm often left wondering **why** certain choices were made instead of others. There is no one right way to do anything, but exploring the options has always been something of tremendous interest to me.

For expression, I usually set aside a few hours at the beginning of the mix process just to experiment with drawing in different expressions directly into DP. I've never done this with Logic, but in DP I can just draw in any random CC curve more easily with a swipe of the mouse. If it doesn't work musically, I can just draw over the old curve without worrying about the "undo" features. (Not that Logic is not as effective in this regard, but I just find DP easier to use, especially with editing and mixing.)

Another thing I've found useful is a trick when trying to make a convincing crescendo with one line:

If I want the violin 1 to crescendo, sometimes it's effective to first ramp down certain other instruments *slightly* for the first part of the crescendo. This tends to be lower instruments which take up important mix space. Sometimes, I'll just ramp down the contrabass a little to make space for the forthcoming crescendo. Depending upon the effect you want, you can can ramp down the other instrument(s) for the first 1/4, 1/3, 1/2.... 3/4 of the duration of the crescendo. Then, begin (or overlap the start of) the violin 1 crescendo. You may also want to ramp the other instruments back up to support the violin1 crescendo as it reaches its peak. The later you start the violin 1 crescendo, the more dramatic the effect will be. The earlier you begin the crescendo, the more subtle it will be.

Often during an overall crescendo, instruments will actually decrescendo over a note or two before continuing the actual crescendo. The resulting curve will look like a rising wavy line. Again, the circumstances dictate the effectiveness of this trick, but when in doubt I always isolate the problem by opening a blank session file to work on it outside of the project session to really nail down details with certain phrase shaping quandaries. I've found that experimenting "too much" within the context of the project psychs me out, and I often find myself starting a project over when I paint myself into a corner.

A composer uses a skethcbook. An artist will do studies and premix paints and test them on a separate palatte or canvas. It seems reasonable to adopt the same "scratchpad" technique for VI.

You mentioned that the perf-leg_sul patches tend to get a bit more like portamento up top with certain intervals. This is true, depending upon where the default controllers are set upon loading them. The cool thing about this is that you've discovered another way to put in an effective portamento without using an actual portamento patch! There WILL be situation where you will need that effect one day. That's why I've been keeping a physical notebook of patch and matrix behavior, organizing it according to articulation. Instead of putting the perf-leg_sul patch only in my notebook's legato list, I'll also put it in the portamento list as a more subtle articulation with this effect. It's not always something that will occur to you at the moment you actually need it, but having a list of patches/matrices with some benchmark controller settings is always nice to refer to days or weeks after you've made the initial discovery. If nothing else, it saves on tedious searching through the samples to re-audition sounds you've already stumbled across.

Onward and upward, eh?

Cheers,
JWL