Legato Problem & Question

ski

Hi DG,

Thank you very much for your reply. I'll gladly post files, but first I'd like to understand what the correct behavior is. I want to rule out my expectations versus the way this function is supposed to operate. Below is an example of the behavior I'm seeing. Please tell me if any aspects of it seem wrong to you...

(my reference for middle C = C4)

[b]14S TrC legato[/b] [i]patch[/i] (loaded into blank matrix)

1a) Play & hold any single note. It will play (loop) indefinitely

1b) Play a note and hold it with sustain pedal. It will loop indefinitely

2a) Play C#4. While holding it, play C#5. You will hear a little bit of "lip" attack on the C#5. This note will sound for a short time and then stop according to the natural performance -- [u]no looping[/u]

2b) Play & hold C#4. Press the sustain pedal and release C#4. It will continue to loop indefinitely. Now play C#5. You will [u]not[/u] hear the "lip" attack on the C#5, and as long as the pedal is pressed the note will sustain indefinitely

My expectation is that the sound heard in step 2B (no attack on the second note, C#5) is what I'd hear by playing as described in step 2A without using sustain pedal. Having said that, can you confirm that all of what I've described is normal behavior for this patch?

Many thanks,

Peter

EDIT: I'd like to add that what I described in step 2A (C#5 not looping) isn't what I expected. Is that normal?

DG

2A and 2B cannot sound the same. 2A uses a start sample and a legato sample; 2B uses two start samples. The is actually the method I use to simulate bowing or legato tonguing.

2A is normal for some instruments, but not all. If you need a looped note, you need to cell xFade immediately after the legato transition to a sustain sample.

DG

herb

@DG said:

If you need a looped note, you need to cell xFade immediately after the legato transition to a sustain sample.

The legato-sus patches (which are placed in the ressource folder for most of the wind instruments) are offering prepared patches, where the interval samples crossfades automatically to the looped sustains.

I recommend to use these patches only for legato movements with long notes.

If the regular legato patch is long enough the results are sonically better using the regular legato patches.

best

Herb

ski

Thank you DG and Herb for your replies. I will reply in more detail shortly, but meanwhile I wanted you to know that I really appreciate your taking time to help me understand this situation. Best, Peter

ski

Dietz, Herb, and DG, again, thanks for your replies.

I've read and re-read Herb and DG's suggestion and explanations, and I've put them into practice on the score I'm currently realizing. I must say, though, that I find the whole "mechanism" for creating legato performances completely counter-intuitive --- particularly the notion that you cannot use legato keyboard technique to produce a legato performance! (Legato, to me, does not include the constant sound of "blips" at the attack of each note except under special circumstances. See below).

In short, my expectations have been dashed. I expected that playing with legato keyboard technique would produce exactly what I hear playing detached notes with sustain pedal (my "2B" above). This is a totally counter-intuitive approach to creating a legato-sounding part. And the idea that I need to switch to a legato-sustain patch to sustain a note? I understand the technique and I appreciate the suggestion. But I must say I don't like the technique at all.

My "2B" method works fine for playing simple legato passages (though it feels wrong to play detached to hear legato), but this method fails marvelously if I want to play certain kinds of basic fast legato passages. For example, using the Bb Clarinet legato patch, play a fast up & down A minor arpeggio -- A3, C4, E4, A4 -- using my 2B method (pressing the sustain pedal after the first note and playing fast detached). Repeating those notes with the sustain pedal down sometimes causes voices to pile up on individual notes. There are additional problems that occur when playing this from the Bb matrix patch, using a controller to switch to a sustained voice. I won't go into details now as this is already becoming a lengthy post.

Now... playing that same arpeggio on the legato patch using legato keyboard technique (no sustain pedal) produces a "watery" sound of "blippy", portamento-ish attacks that are absolutely charming for polkas, Klezmer, even occasionally passages in "serious" classical music. But an accomplished clarinetist playing "serious" music would strive [i]not[/i] to sound so watery when playing legato, even if the passages were fast and difficult. Sure, playing fast passages and hearing those "blips" implies that a passage is difficult to play. So in many cases the "blips" are more of a "special effect" than something which wants to be heard all the time.

Dietz, I appreciate the tremendous effort that went into the development of the legato patches. I would respectfully suggest that now it's time to devise an algorithm that will allow for normal [u]legato[/u] playing technique that would create the effect I described in "2B", with the sustain pedal used to indicate that subsequent notes are to be played from the sustain portion of the sample, not the "blips" of the attacks of the legato samples. And further to that idea, the "blippy" intervals should available as a performance option that can be switched on/off as desired. Like I said, it's a good special effect.

Anyway, I'm sure that's enough from me by now...

DG

Peter, I still think that you are not quite understanding the legato function. The whole point is that you do play them legato, in the same way as you do on a keyboard. The fact that you can't do the same when using the sustain pedal is a huge plus, not a minus. It is the only way to force a start note, whilst playing legato. May I also respectfully suggest that using the sustain pedal is not a normal legato technique; it is a sustain technique that is often used because of bad keyboard skills. Not saying anything about yours though.[;)]

I still suggest that you post a MIDI file containing those "blips". I only get those when the legato note runs out, and often in these cases (such as with solo strings) the transition material heard with the legato would be totally inappropriate, as a bow is only a certain length and would need to change anyway. I agree that if all legato notes were looped it would make things easier, but it very rarely causes me problems.

However, if you find that you hear too much transition material between notes, then just raise the release time, so that it is partially obscured.

DG

herb

It's a little difficult to understand your problem without any audio reference.

I just can post another example, this time the clarinet also using extensively performance legato patches.

http://vsl.co.at/Player2.aspx?Lang=1&DemoID=4650

All this solo instrument demos were performed live on a keyboard without any editing.

You are right that there are differences between a slow and a fast legato movement,

that's why we offer normal and fast legato performances (and also the performance trill patches, which also work very well for fast legato performances)

Also our speed controller helps to automate the switch between slow and fast passages

Just another few things which are important to know:

- perf-legato patches are monophonic,

- you have to perform legato (there shouldn't be breaks between the notes)

- it could be tricky if you perform a legato line changing from piano to forte (during legato)

here the transition between the notes gets obvious

In this case it is better to activate velocity crossfade, and make a fast crossfade change from low to high.

best

Herb

Dietz

A question to Peter, just to make sure (as I saw him posting a similar topic in another forum, too) : We are talking about the Vienna Instruments, _not_ the legacy Horizon Series products of the Vienna Symphonic Library, which have to rely on EXS24, GigaStudio, etc. - is this correct?

/Dietz

ski

...see next post...

mahag

Yes, Peter, if I may jump in. I hear what you are calling the "blip" although it's very subtle. In the clarinet example I think it's just the subtle mechanics of the instrument being heard, the slapping shut of a pad, etc, etc. I'm quite sure it's only audible in the starkest of situations (solo and little reverb maybe) and because you've become accustomed to hearing that ultra smooth legato, though not as realistic, these "blips" are catching your ultra sensitive ears. I've noticed it once or twice before also, but in the context of a musical piece it always sounds great.

ski

Aha! In the score I'm working on now there was a small clarinet part, all of 4 notes, and this demonstrates "the problem" perfectly.

First, an .mp3:

http://www.score2picture.com/vsl/clarinet.mp3

What you'll hear are four notes played legato, no pedal, just some modwheel--->velocity x-fade to shape the phrase

For anyone interested, below is a link to a .mid file of the part (the part happens around measure 24). Someone above requested an .fxp file. Sorry, but I have no idea what that is... I saved a .matrix file of my custom programming on the stock Bb clarinet, don't know if that would help. But you should be able to use the stock Bb Clarinet that comes with VI Special Edition, activate the legato keyswitch, and get the same results.

http://www.score2picture.com/vsl/clarinet.mid

Now, here's my commentary about what I'm hearing in the above .mp3 (all pitches concert):

1. D to A
If this were played on a real clarinet you would have to 'cross the register break'. Clarinetists work very hard to play across the break smoothly and with even tone. I'm sure you guys know this. And as an amateur clarinetist myself I know the difficulty. But if we're going for a professional sound here, there should be no "bump" between these notes. But indeed you'll hear it in this legato patch!

2. A to B
Now that the break has been crossed, playing these two notes on a real clarinet should produce very little if any break in the tone. Heck, even I can play those notes legato and seamlessly on my plastic Bundy with a worn-out reed. But here we hear a "bump". It sounds almost tongued. So for me, going from A to B is not a legato sound. It's smooth, but it's not legato.

3. B to A
The little bump between notes is OK for me. It's not objectionable, and adds realism.

But back to my main point --- playing the simplest of lines (no wide leaps) using standard legato keyboard technique results in "bumps" between notes in legato patches. Clarinet. Bassoon. Trumpet. Flute, and so on. It's just not the sound I expect when I think of legato.

Anyway, it's super late now so I'm going to sign off for now.

strytten

Peter's questions capture the sort of questions I've been accumulating as I've been working my way up the learning curve. I've done the video tutorials, and I visit the user area (but there are so many postings, it is time consuming to find relevant postings even with the search feature). I am using the velocity xfade feature. It would be really helpful for newbies like me if others could post any techniques they're willing to share to get more realistic legatos (or other types of sounds). Thanks! Steve Trytten, Pasadena, CA

ski

I'm finding it extremely difficult to play simple legato lines with just about any legato patch/matrix I call up. Regardless of how much custom programming I do (or x-fading to sustain samples, as Herb suggested above) I'm finding something inherently problematic in the way the legato function is implemented. It might have to do with the samples themselves, or how their attacks have been trimmed (or not). To achieve legato with a consistent sound isn't intuitive. But in so many cases there is a pronounced attack on legato samples.

I'm working with the stock french horn right now. With the legato samples selected (low D#), play middle C up to middle G (legato of course). The C has a nice attack quality, but then when you play the G you'll hear an attack on the G! That's not legato! Maybe it's a kind of tongued legato, but it's not legato as I would normally associate the term with the sound. (There's also a lousy pitch bend on that G, but that's another matter).

Now, try it a different way. Same notes, but before you play the G, hit the low C# to access the sustain samples (not the legato-sustain). Now, [i]that's[/i] the sound I'm after. That's the sound that a skilled player would be able to achieve. But the legato patch by itself doesn't provide that kind of sound.

Above, mahag suggested that maybe my ears were sensitive and that I'm picking up on these kinds of anomalies. I respectfully disagree. It's not a matter of my ears being sensitive, or even over-sensitive. It's a matter of knowing what the real instruments sound like and my not hearing that.

So what kind of sound do I associate with the term "legato"? Well, it's not "tongued legato". I'll give two examples of what legato means to me:

1. VIOLIN: When playing a legato line on one bow stroke (and on one string), the player essentially does what a guitar player does -- they either hammer on, lift off, or slide to change pitches. On a single bow there would be little re-attacking of new notes, if any. The pitch wouldn't change as though someone flipped a switch, but it would be pretty close to that.

2. CLARINET (again): On a single breath, and without re-tonguing notes, the clarinetist puts fingers down or lifts fingers up to change pitch. Sometimes the pads will not lift or close quickly enough, and sometimes a finger won't be placed or lifted on an open hole quickly enough. In those cases it will produce a bit of a "re-attack blip" on the sound. But played by a skilled clarinetist -- the kind of thing I'm trying to simulate -- you wouldn't hear too may of those "blips".

So that's the kind of sound I expected from "legato". I think the implementation of VSL's legato is brilliant in concept. And sometimes those little blips/re-attacks sound totally convincing. But for the purpose of creating simple fingered legato lines (on a single breath or single bow, depending on the instrument), it's a difficult task at best. What's worse is that there's no consistency within the respective instruments. Some connected notes sound smooth, some don't.

So as I said earlier, this whole thing may be a matter of what I expected "legato" to sound like using this library versus what it actually [i]does[/i] sound like. But at the same time I don't think my expectation is too off-the-mark. Yes, I'm disappointed, and IMHO I see lots of room for improvement here. But I'll add that overall I'm not disappointed with having bought into VSL. It's serving me well all the same.

brian_30027

Just to chime in belatedly, having seen this thread for the first time...

I'm in agreement with what Peter Schwarz has argued. As a clarinetist with professional-level training and with some years of experience playing in orchestra from my university days, I know a little something about how the instruments should sound. The VSL legato instruments (at least in the Special Edition, which I have) do not always acheive a normal, natural-sounding result (clarinet incidentally being among the worst IMO, but then some other instruments sound quite good). This has been one of the few disappointments I've had with VSL (but a serious one). When the transitions don't work well, the effect is like hearing amateur players with bobbling, ill-coordinated fingers. I hope that this can be improved and fine-tuned in a future update, because I've often had to expend quite a bit of extra effort, using techiques similar to Peter's, in order to work around the problems when they are bad enough to interfere audibly with the whole mix. Additionally, two other problems I have had with the legato instruments is the lack of strict playing response (i.e., the legato transitions still activate when there is a small gap between the midi notes, making it difficult to simulate tenuto- or legato- tonguing) as well as the difficulty in getting a clean, firm attack on the first note of a phrase with some of the winds.

ski

Brian,

Thanks for chiming in. I'm really glad I'm not the only one who feels this way.

Until now I've been willing to concede that my disappointment in the legato patches might be a matter of my expectations not being met, i.e., my idea of legato may simply not the same as those people who programmed VSL. But... after using VSL for 3 weeks now (and trying my damndest to find realistic and consistent workarounds for the legato glitching) I'm no longer willing to consider a justification for it.

Speaking generally, the meaning of what a "legato" sound is (on any instrument) is not, IMO, really open to debate to any great degree. Legato playing is a technique that musicians learn to achieve very early in their training. At least amongst classical musicians I think it's fair to say that there's a fundamental understanding of what "legato" sounds like on any particular instrument. And in general the VSL legato patches do not achieve this paradigm [u]out of the box[/u], and simply by playing the legato patches themselves. Heck, it's virtually impossible -- if not [i]actually[/i] impossible -- to achieve a legato sound even between two adjacent notes (whole or half step) on so many of the VSL legato instruments!

But there is some hope... After discussing this subject on another forum, I've learned from a chap named Daryl an approach to making VSL legato patches actually sound legato. The technique involves routing your sound through a reverb (small room) and listening only to the wet signal. The room sound smoothes out the bumps in the legato patches.

Try this:

Select the chamber violin legato patch -- dry, no reverb at first -- and play some legato phrases (at varying speeds) using only the notes G2 through C#3 --- this representing notes which can only be produced on this single string and don't extend past the first position on the string. Imagine playing the part using long, continuous bow strokes. I guarantee you that some of what you play will likely sound just awful (not your music, the legato transitions!). Next, try this technique of routing that patch into a small room and listen only to the wet signal. Assuming you've chosen an appropriate room (which might take some time to find), there's a good chance that the bumpy legato transitions will no longer be an issue! Those same lines that sounded really crappy and stilted before will now sound musical!

The downside to this technique is that you cannot use VSL legato patches to produce intimate sounding, close-mic'd instrument simulations. Utilizing a reverb for this purpose commits you to a room sound for that instrument, if not the rest of the instruments in your ensemble. Depending on what you're doing, this may or may not be desirable

I realize this is not a new technique -- listening to a sound as though it were in the context of a room and the listener at a distance -- but it's one that seems to be necessary in order to make legato patches work.

Anyway, that's all for now.

brian_30027

Peter, Thanks for sharing the tip. I had noticed that reverb used in the regular way tends to improve things, but this technique will be an additional help in some situations. I think you are right to feel that this is not entirely a subjective matter. Let me put it this way -- I can remember taking auditions in which I was asked to play specific scales or intervals just to demonstrate that I could play a smooth and effortless legato. If my playing had sounded like the VSL legato clarinet does much of the time, I would have failed that part of the audition, no doubt about it! Of course there is no way that the technique used by VSL can ever be perfectly realistic (to get close to that, you'd need the engine to have some kind of real-time spectral morphing inbetween samples). But I'm optimistic that it could be greatly improved to the point of being reliably *musical*, if they put some sufficient effort into fine-tuning. After careful listening, it seems to me that the majority of bad-sounding legato intervals are caused by the transition samples either being too loud (given equal-velocity notes), or slightly too long and thus too conspicuous. A few, however, seem to be the result of less-than-ideal performances. Even without an update to the samples, perhaps an elegant partial solution would be for VSL to add a new feature where you could use a controller to switch on the fly from their legato mode to a conventional crossfade-and-skip-the-attack mode, where desired. Anyway, that's how I see it for now.

ski

A hearty "Indeed!" to everything you wrote. And I do mean "everything".

I agree, there are some bad performances. It seems to me that on some instruments, and for some notes, the repetition sample is the better of the two.

Being able to crossfade from one sample to another really isn't that hard to accomplish provided the raw material (i.e., the sampled notes) are adequately performed. But even with less-than-perfect samples, if the idea of connecting notes in legato fashion is tantamount to crossfading between the sustain portions of those samples, the attack of such a "bad sample" can often be entirely overlooked. There are much less expensive sample plugs on the market (which shall remain nameless) that accomplish this quite successfully. While there is no comparison when it comes to the difference in sound quality between them and VSL, as it is, many of the legato VSL patches can't be used for their intended purpose anyway. And by "intended purpose" I mean the universal understanding of what legato sounds like, per my above post.

So there's lots of room for improvement here. Meanwhile I use VSL for the strings and brass and sustained woodwind patches (the clarinet sustain in particular having received high compliments on its sound quality from the composer I'm currently working for).

Yin and Yang here. Wish there was more of (whichever is the good "Y" word) with VSL legato patches.

Sincerely, Peter

Products

Deals

Music

Forum

Support

Academy

Vienna Symphonic Library Forum

Forum Statistics