• Register
  • Help
Page 1 of 2 12 LastLast
Results 1 to 10 of 11

Topic: On the Virtual Stradivari and Gofriller approach to Vibrato and related issues.

Share/Bookmark
  1. #1

    Post On the Virtual Stradivari and Gofriller approach to Vibrato and related issues.

    Hi all,

    The recent thread on the “Morphing Gofriller” in this Forum induced me to propose some basic considerations on vibrato of string instruments, either real or virtual. Please forgive me if they are too obvious.

    In the real instrument, vibrato stems from pitch changes, related to cyclical shortening/lengthening of the string. Those frequency oscillations are converted into amplitude oscillations by modal resonances occurring within the body of the instrument. These amplitude oscillations are typically out of phase for the different harmonics. The latter fact explains why a convincing vibrato cannot be obtained by applying a combination of pitch and amplitude LFO to senza-vibrato samples.

    Re-synthesis of vibrato in the Stradivari and Gofriller, is based on a more sophisticated approach, aimed at reproducing the behaviour of the real instrument. In essence, LFO-generated frequency oscillations are fed to a convolver, loaded with the modal resonances of the body of the instrument. This recreates the out-of-phase amplitude oscillations of true vibrato.

    This approach yields several advantages. First, the harmonic structure of the re-synthesized vibrato can be very close to the original con-vibrato sample. Second, both the intensity and rate of vibrato are under player’s control. We should focus on this latter point for a better understanding of the potential limitations of this approach.

    In real life, the player performing vibrato applies a pitch oscillation which is time-varying in shape, intensity, rate, and low-frequency content. The virtual instrument basically uses a fixed-shape modulating waveform, which can be applied at variable intensity and rate by acting on two midi controllers. Low frequency pitch modulations can also be controlled by the pitchwheel.

    All this means that, in order to mimick the original behaviour of real vibrato, the player needs to introduce these time-varying pitch modulations by simultaneously acting on three midi controllers.
    Omitting this step will lead to kind of a static, lifeless vibrato.

    While this time-varying pitch modulation can be effectively carried out by humans using the proper midi setup (modwheel, aftertouch, pitchwheel), this approach is clearly more complex than just using built-in vibrato samples. It involves a learning curve, a bit of experimentation, and some knowledge of the playing technique of the specific instrument.

    Thus, it is not surprising that re-synthesized vibrato may at times sound less than satisfactory. This is more likely related to a yet incomplete understanding of the most effective pitch modulation, than to conceptual or structural defects of the instrument.

    These instruments still need to be fully explored to exploit their true potentialities. That’s why we rely on constructive feedback to work out further improvements.


    Sincerely,

    Giorgio

  2. #2
    Senior Member
    Join Date
    Dec 2004
    Location
    San Francisco
    Posts
    150

    Re: On the Virtual Stradivari and Gofriller approach to Vibrato and related issues.

    Quote Originally Posted by Giorgio Tommasini
    All this means that, in order to mimick the original behaviour of real vibrato, the player needs to introduce these time-varying pitch modulations by simultaneously acting on three midi controllers.
    Omitting this step will lead to kind of a static, lifeless vibrato.

    Thanks Giorgio. Well-explained.

    Might it be feasable to introduce sufficient "noise" (i.e., random rate of periodic drift/correction around a user-input target value, which rate varies proportionally to that target value) to mimick the real behavior, such that the instrument automatically introduces intensity/rate/pitch drift at a realistic level, and the user need only employ the modulation wheel to control vibrato?

    With regard to the overall "tone" issue, wherein the Strad emulation is superior to the Gofrilller in its current form: comparing Synful's difficulty with the realistic emulation of bass clef instruments generally is strongly suggestive of the limitations upon realism that phase-alignment imposes on time-varying spectra in the low registers, both in the context of RPM (Synful) and of pre-rendered alignment of bitstreamed samples (Garritan, i assume).

    Might the introduction of noise, in a manner similar overall to that described above, improve the realism of phase-aligned low register instruments? This would introduce some additional CPU overhead to such instruments; bass clef instruments need to be "smarter" to be realistic because an additional characteristic of the sound will be derived synthetically. An improvement in tone might be worth it however.

  3. #3

    Re: On the Virtual Stradivari and Gofriller approach to Vibrato and related issues.

    Jloeb,

    This, as well as other approaches to semi-automatic modulation, are under investigation.

    The drawback, as you can figure out, is that any automatized approach potentially limits player's control of expressiveness.

    Phase-alignment per se does not necessarily yield timbrical differences with respect to the original samples. Synful, as far as I know, relies on additive synthesis. Thus, eventual timbrical issues do not share common causes.

    Giorgio

  4. #4
    Senior Member
    Join Date
    Dec 2004
    Location
    San Francisco
    Posts
    150

    Re: On the Virtual Stradivari and Gofriller approach to Vibrato and related issues.

    Quote Originally Posted by Giorgio Tommasini
    The drawback, as you can figure out, is that any automatized approach potentially limits player's control of expressiveness.
    Yes. Linking intensity/rate drift to mod wheel and pitch drift to volume expression could produce a two-controller result that might be acceptable and intuitive, mimicking some limitations which the real player encounters.


    Quote Originally Posted by Giorgio Tommasini
    Phase-alignment per se does not necessarily yield timbrical differences with respect to the original samples. Synful, as far as I know, relies on additive synthesis. Thus, eventual timbrical issues do not share common causes.

    I'm not certain. While the sound-generation modes are different, the common cause may be greater Hz proximity of the fundamental to the dominant overtones in low register instruments. The increased psychoacoustic effect of aligning phase would be geometric relative to decreasing pitch frequency (i think). Subtle, since the rates are still pretty far apart, but it seems the human ear has a way of picking this up; thus perhaps the tone difficulties in p.a.-bass clef instruments, and why synthetic drift-noise might help.

  5. #5

    Re: On the Virtual Stradivari and Gofriller approach to Vibrato and related issues.

    Thanks, Giorgio, for your explanation. Makes all sense.

  6. #6

    Re: On the Virtual Stradivari and Gofriller approach to Vibrato and related issues.

    The nice thing about samples is that you have real vibrato applied by a professional player. The bad thing is that you can't change it.

    That leads us to pre-convolution vibrato. Garritan's solo instruments and Giga Violin both offer such a solution. The problem now lies in the hand of the midi performer. If you apply vibrato well, it can sound very realistic. If poorly performed, you have the chance to sound like a poor string player.

    That's always the problem with more power and control. They also bring more responsibility.

    The question for the midi performer: are you up to the challenge?

    (BTW, since December I have a real violin at home. The most difficult aspect of the instrument? Good vibrato!)

  7. #7

    Re: On the Virtual Stradivari and Gofriller approach to Vibrato and related issues.

    And, i should imagine, this self choice approach will enable us to perform an even more realistic vibrato, given practise.
    One step closer to a 'live' sound that we can apply individually to a piece based on the nuance of the piece. Only speaking for myself, i like the idea of being able to more completely control this aspect of string sound production. Then it's down to me how it comes out.

    A step forward in my view.

    Good work to the team that put this together.

    Alex.

  8. #8

    Lightbulb Re: On the Virtual Stradivari and Gofriller approach to Vibrato and related issues.

    I think that some feeling is honest, but less effective, when it's based on the memory: we compare the sample with an ideal concept we have in our mind, and we are frequently frustrated or not satisfied by the difference.

    To make easier talking about vibrato type, speed, intensity and variation, I collected several samples of real players, and pasted the beginning in sequence, including some artificial samples.

    2 are me, one is Yo Yo Ma.

    The rest is Internet Surfing (do it yourself, if you want: it's interesting... ).

    THE SWANS PARADE MP3

    THE SWANS PARADE AIFF

  9. #9

    Re: On the Virtual Stradivari and Gofriller approach to Vibrato and related issues.

    Quote Originally Posted by JonFairhurst
    That's always the problem with more power and control. They also bring more responsibility.

    The question for the midi performer: are you up to the challenge?
    Frankly, no! I used the Strad in a song and thought I'd created a passable violin performance. When it came to the mix I realised it was unusable.

    You can't just use the Strad naievely. You have to invest time mastering it as an instrument in it's own right, but before that you need to understand vibrato.

    Perhaps the best solution to working with this software instrument is to go and buy a violin, and then come beack to it after a few years. Otherwise a study of vibrato is essential.

  10. #10

    Re: On the Virtual Stradivari and Gofriller approach to Vibrato and related issues.

    Quote Originally Posted by hywyn
    Frankly, no! I used the Strad in a song and thought I'd created a passable violin performance. When it came to the mix I realised it was unusable.

    You can't just use the Strad naievely. You have to invest time mastering it as an instrument in it's own right, but before that you need to understand vibrato.

    Perhaps the best solution to working with this software instrument is to go and buy a violin, and then come beack to it after a few years. Otherwise a study of vibrato is essential.
    Or, practise with the one you have, Hywyn.
    Despite the fact that these instruments are based in a computer, learning to use them is no different to learning to play one 'live.'
    It's just another instrument to master, and explore possibilities with.

    Regards,

    Alex.

Go Back to forum

Bookmarks

Posting Permissions

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