• Register
  • Help
Page 1 of 6 1234 ... LastLast
Results 1 to 10 of 54

Topic: Legato!

  1. #1



    a simple word but a huge challenge for the sampling world.

    You all know these pumping performances with swelling attacks where the envelopes try to catch up with the next note - without success.

    Or if you cut away the attacks in order to have a quicker response: Liveless generic sounds, fast responding, but not LEGATO.

    As long as we operate with single samples, legato is a dream. Until the whole stuff is getting really virtual we are to use special techniques to get a near-legato feeling.

    I am interested to share some ideas and experiences with you about how you are getting near-legato today, using GS.

    Any experiences and suggestions are very welcome.


  2. #2

    Re: Legato!

    Check out the Gigastrings first demos I would swear Garritan got his hands on something there.
    Can\'t wait to put mines on his lib.

  3. #3

    Re: Legato!

    I think Gigastrings are working well for slow layers, and for staccatos and spiccatos in a medium tempo, but the legato examples didn´t impress me.

  4. #4

    Re: Legato!

    Well I\'m gonna get those strings asap and I\'ll tell you then who \'s right or wrong ?!?
    I really think I heard something promisingon the legato side so I\'ll take a chance.

  5. #5

    Re: Legato!

    I would love to hear some legato samples of the Gigastrings violins as well. That is the area that most concerned me in the original slate of demos.

    I believe Gary said he was addressing that problem specifically.

  6. #6

    Re: Legato!

    I don\'t want this to be considred hype its just an honest opinion from working with gigastrings

    The legato feature is still being tuned, but what I\'ve used of it is really good IMO. Its pretty flexible and be adjusted on the fly. We are all still learning how to use it, and what works and what doesn\'t. Still with the legato feature you can remove the \"pumping\" that occurs with constantly triggering samples with long attacks. As well, since it can be controlled \"on the fly\", you can create a more lively sound rather than using samples without an attack. Its a promising and unique feature

    You can\'t create long realistic fast trills with this feature (tho there are trill samples and some of the shorter articulations can simulate some), but for moving melodies it can be extremely useful, adn with adding touches of slides here and there can be extremely realistic. ... the word is realistic...not real...

    As for creating it with other libraries.

    With AO the \"legato\" key switch can be manipulated into sounded less abrupt with some mod wheel controla and expression control. It takes ALOT of work to do tho and is usually not the most gratifying result. Layering different strings together and using their \"legato\" attacks if they have any helps. Synth strings with \"legato\" mode over sampled strings help as well. Most of this isn\'t the most realistic way of working, but it can help ease the swelling/pumping sound. It all takes a bit of work tho

    Really...I am an Idiot

  7. #7

    Re: Legato!

    <BLOCKQUOTE><font size=\"1\" face=\"Verdana, Arial\">quote:</font><HR>Synth strings with \"legato\" mode over sampled strings help as well. <HR></BLOCKQUOTE>

    Yeah, this is what I end up doing now, and I\'d love to have GigaStrings let me stop wasting time trying to make my strings sound legato.

  8. #8

    Re: Legato!

    I cannot imagine in theory how you can achieve legato sound without cutting attacks: The problem is, that it takes some time after the bowstrike to gain full volume. In real life legato is achieved through moving the bow while changing the tune. There is also a very short portamento effect. You can hardly simulate this with samples.

    What I tried with some success is to cut off the very start of the attack and to add a very quick artifical attack in order to have no clicks. Then I use a combination of Pre-Attack Level of e.g. 20% to 30% together with an attack time of about 2 to 3 seconds, dependant on the character of the sample.
    That makes quite a nice \"cantabile\" sound. I also use a controller to steer other combinations of Pre-Attack and attack time for slower or faster passages.

    It has payed to not cut off attacks with ff-samples, because that spoils authenticity and makes a flexible but cold \"generic\" sound. Her you might want to use the sample-start feature in combination with pre-attack level and attack time and to steer
    these with a controller through dimensions. I wished that sample-start would have a wider range than just 2000 samples.

    I never thought of layering. Could anybody tell me, what you layer with what and how you do it (timing parameters etc.) ?


  9. #9

    Re: Legato!

    >>>I cannot imagine in theory how you can achieve legato sound without cutting attacks: The problem is, that it takes some time after the bowstrike to gain full volume. In real life legato is achieved through moving the bow while changing the tune. There is also a very short portamento effect. You can hardly simulate this with samples.<<<

    Actually, Its something thats in Garritan\'s Orchestral Strings. It can be done (including the slight portamento), and we are all trying to find ways to make it easiest to do. The cellos sound amazing with this technique, so it can be done with samples.....and before everyone goes asking where the demos are, they\'ll pop up soon...but not till they are ready to be posted.

    As for layering techniques, there are all kinds of things one can do. I find that filtering out some lows and low mids on some samples give me a smaller section sound that you can \"simulate\" divisi strings with.

    Some synths (my Roland XP does) have \"legato\" mode that you can use layered under teh samples, but it still sounds like caca sometimes...then sometimes its not so bad..

    As for other instruments, I\'m sure with some sample editign I can get some uniue sounds out of Dan Dean solo WW. Playing legato isn\'t too bad as they are...actually they are pretty good, I wish there were \"key pressing\" samples layering those on top sublty would be a GREAT legato simulater.

    Scarbee seems to have nailed it with his library....

    Horns... doing some pitch bending before each note helps for legato on frech horns..but its VERY difficult to do in real time..and is sometimes more work than its worth

    Trombones...someone should listen to Scarbees bass lib and we might actually get some cool trombone slides and acctually be able to recreate some....until then we have to settle for......<cringe> pitch shifting

    Melodyne is out for the Mac and I may look into that for recreating slides...dont use my mac much for audio tho...maybe its time to get a titanium hehee

    Trumpts.. I need to go in and edit the ones I have for \"legato\" I think it can be done tho .... when I get around to it... which will probably be never...

    Really...I am an Idiot

    [This message has been edited by KingIdiot (edited 08-09-2001).]

    [This message has been edited by KingIdiot (edited 08-09-2001).]

  10. #10

    Re: Legato!

    So that we have a common ground for discussion, indulge me for a moment as I summarize the “legato problem” and then describe some possible solutions. I will focus my comments on string sample libraries, but what I have to say here should apply to woodwind and brass sample libraries.

    An inherent problem with string libraries, at least when faced with the task of playing a legato line, is that the sampled tones are discrete and contain the onset of a bow stroke. For some musical contexts, these samples are desirable. However, when using the samples to create a melodic line, most if not all of us perceive the transitions from tone to tone as stilted or “drunken,” particularly if the samples contain a noticeable swell. Some forum members have already suggested the use of the “overlapping” technique as a means for concealing the effect, but the technique doesn’t always work and is a poor substitute for the continuous bowing “information” that is essential to the sound of a legato phrase.

    As we all know, “legato” is a playing technique whereby the string player smoothly articulates two or more tones under a single bow stroke (a change of bow direction may be required for longer passages, but a well-trained violinist will conceal the transition). Only the initial tone of the legato phrase contains the onset of the bow stroke; the other tones result from depressing the already vibrating string(s) at the appropriate locations on the fingerboard. Of course, as string players will attest, much more is involved here (slides between tones, continuity of finger and hand movements, pressure adjustments of the bow, etc.), but for the sake of moving this along, I’ll end the description here.

    One solution to the legato problem, as I have described it, is to sample every conceivable legato phrase in every conceivable tonal and non-tonal context. OK, not feasible! Moving on. Another solution is to sample every note within an instrument’s full (professional) range in the context of a legato phrase. The developer wouldn’t have to slice off the attacks because the samples wouldn’t contain information specific to the onset of a bow stroke (I’m borrowing the high-falutin expression “information specific to . . .” from a group of researchers, known as “ecological psychologists,” who study perception in terms of the information in sound and other “ambient arrays,” information specific to its source and the source’s transformation). Unfortunately, the acoustic information about the smooth fingering transitions between tones may still be missing, to say nothing about how many samples the developer would have at the end of a recording session (samples with bow attacks, samples without bow attacks, etc.). In other words, this, too, may not be a workable solution.

    So, this takes us back to where we started. (I’m not much help, now am I.) I propose that, instead of returning to the concert hall or recording studio to re-record everything, developers should grapple with what they already have: discrete tones with conspicuous bow attacks.

    As a beta tester for Gary Garritan’s “Orchestral Strings,” I’ve witnessed just this. At the urging of several beta testers, Gary, Tom and a “clever” beta tester dealt head on with the legato problem by answering these two questions (two sides of the same coin, as it were):

    1. How do we simulate the continuous bowing information specific to legato playing?
    2. How do we conceal the realistic, but disruptive bow attacks?

    Without going into any detail, I can say that their answer to these questions relies on specially designed .gig instruments and the use of the “Legato Mode” feature in a third-party utility called MaestroTools. I hasten to add that Gary’s team confronted the problem early in the beta program and spent countless hours brain-storming and approaching the problem from several different angles before arriving at the present and very flexible solution. (BTW, one approach involved lopping off the attacks, but they soon concluded that the result was unconvincing and then moved on to the next experiment.)

    I’ve used the Legato Mode feature for several weeks now and am quite satisfied with the results. I frequently play phrases with and without Legato Mode enabled to hear the “before-and-after” effect, which is most revealing. When Legato Mode is enabled, the transitions between tones are much smoother and well balanced. As I play a group of notes, I even sense the bow being drawn across the strings; granted, this may be my imagination playing tricks on me, but the “auditory image” is very much real to me. Of course, practice is required; the more time I spend with Legato Mode, the easier it becomes to produce realistic and elegant sounding legato phrases.

    Will Legato Mode satisfy everyone? I hope so (as does Gary, no doubt), but I recognize that no two auditory systems are alike. Some of you, particularly string players, may crave even more acoustic information in the samples or may pick up information that is distracting, since you are attuned to acoustic properties that the rest of us either ignore or are not trained to detect. Moreover, the verisimilitude of perception is not always “pure”: you will hear what you want or expect to hear, due in part to your predispositions and the anticipatory nature of perception and cognition. In short, realism is relative.

    Alas, developers will just have to live with the reality that not everyone will be pleased with their work. A short visit on this forum is enough to convince me that they’re doing that already. Poor, wonderful souls!


    [This message has been edited by PatS (edited 08-13-2001).]

Go Back to forum
Page 1 of 6 1234 ... LastLast

Tags for this Thread


Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts