• Register
  • Help
Page 1 of 3 123 LastLast
Results 1 to 10 of 28

Topic: Variable string section size / harmonics

Share/Bookmark
  1. #1

    Variable string section size / harmonics

    I'm working on a recording of quite a complex contemporary score, where the degree to which the string sections are divided varies considerably, and I'm wondering if anyone has any tips on how to distinguish between them most convincingly.

    The strings are 12 each of first and second violins, 10 each of violas and cellos, and 8 basses. I've broken down the way they divide into separate tracks as follows:

    Full section
    Large div: about 2/3 of the section (sometimes all the violins together are divided into three lots of 8)
    Div: half the section, as normal
    Medium div: rather less than half
    Small div: as few as three players (this only applies to violins and violas)
    Solo: there are a number of passages where one or more sections are fully divided into solo parts - a couple with all 52!

    Naturally, the solo parts are easy: use the solo instruments. In a few places there are just two players on a line, so I'm doubling up solo parts, but beyond that I'm not sure how best to approach the problem.

    I thought of basing it all on the div (half) section and somehow beefing it up a bit for the large div and rather more for the full section, then somehow (how? EQ?) thinning it out for the smaller divs.

    For the beefing up I've tried adding subtle chorus or delay, but whatever I do sounds like an 'effect', which of course I don't want. I've also tried doubling up for the full sections, moving the duplicate tracks a few ticks late and panning them slightly off from the originals, but even that sounds like an effect.

    Maybe I should just do the whole lot as solo parts, multiplied up as need be, with a good dose of 'humanization' and some intonation and timbre variation?

    Oh... and what do I do about harmonics? I haven't tried it yet, but I thought of pitch-shift and a bit of subtle EQ...?

    For what it's worth, I'm using lovely new Digital Performer 5.1, but I don't have a budget for expensive plug-ins beyond what I've got with DP.

  2. #2

    Re: Variable string section size / harmonics

    Hi,

    are you speaking of GPO?

    There are no half sections or such in GPO. A simple workaround would be to use polyphony within one group and reduce the volume by cc#01.

    Maybe it is not the case with your piece but it occurs to me that people in the samples scene take the divisi topic much too serious. Orchestra players take care that every voice will be heard but will not make a big secret about it. So they either will play a passage with real divisi or double stops. With triple stops it is common that the first group plays the upper two notes as a double stop and the second group plays the lower two. Which will result in a thicker middle voice of course.

    Then you never know whether the players will compensate the volume because they know that it is divisi now.

    I personally tend to put a second voice in there sometimes, and that's about it. Mostly real players would play this in double stops, and double stops are not necessarily softer per voice than a single voice, often louder. So adding a second voice is OK for me in the most cases.

    There are concepts that make a science of it and it may make some sense. Synful has an automatic divisi, and there is a new high-prized library around the corner that also has one. But it is not THE big secret to a good strings sound, how quite some marketing people try to convince us.

    Just an unspecific rant, maybe totally out of place with your piece. In that case, sorry, hehe.


    Hannes
    All your strings belong to me!
    www.strings-on-demand.com

  3. #3

    Re: Variable string section size / harmonics

    Hi Hannes

    Thanks for the instant response! However...

    I am indeed speaking of GPO, and of course I'm aware there are no half-sections.

    The polyphony idea won't work for me I'm afraid, although volume isn't an issue, as I plan to record each batch (full, div, soli, etc.) separately to audio, then I can mix the audio tracks accordingly. Within reason, I don't mind how many audio tracks I end up with.

    Maybe it is not the case with your piece but it occurs to me that people in the samples scene take the divisi topic much too serious. Orchestra players take care that every voice will be heard but will not make a big secret about it. So they either will play a passage with real divisi or double stops. With triple stops it is common that the first group plays the upper two notes as a double stop and the second group plays the lower two. Which will result in a thicker middle voice of course.
    Well, I really would not want them to do that with my piece! The purpose of divisi for me isn't simply to get more voices - it's a different sound: lusher because of the denser harmony, but at the same time thinner because of the smaller number of players on each voice.

    In any case, how do I have the players play two different glissandi simultaneously? By which I mean starting and finishing at different times, and overlapping. Ever heard any Xenakis? That's an extreme version of what I'm talking about.

    So yes, I'm afraid your 'unspecific rant' is out of place with my piece, Hannes, but thanks again for the swift reply!

  4. #4

    Re: Variable string section size / harmonics

    Well...

    I understand what you want to do, and I think that the best approach is to divide all strings to solo and start playing!

    I'm not familiar with GPO, but indeed if you are anyways going to have solo strings (52, ala pederevski) then you might as well, copy the parts that they play together in all tracks and alter slightly the velocities and move them a few miliseconds each to get the sound unison and then solo and divisi a3,4 etc...

    Not sure if EQ would help...

    About harmonics I had an idea about band pass (which would take out, all the bass frequency and leave only the harmonic) and mixing it with the real note, somehow to reach a certain level... pitch shift is a bad idea for a whole octave I'm afraid.

    Any chance of seeing the score, just in case there is a way out? Mostly because the work sounds really interesting...

  5. #5

    Re: Variable string section size / harmonics

    With GPO you can use the string sections, the solo, or the 9 "Plr" voices of course, but there are limitations.

    You will get inconsistent results if you use the same instrument for different lines whereever unisons might occur. Unless you are certain that there will be no unisons between lines, you must distribute the 9 "Plr" instruments among the separate lines, 4/5 for two, for example, or 3/3/3 for three, etc.

    If the samples are triggered exactly simultaneously on unisons, the apparent volume of the duplicated sample will simply double without creating a natural texture of two instruments. If not simultaneous, you will hear unnatural phase cancellation (I think you mentioned this). You will have this difficulty if you intermix the "Plr" voices with their respective solo voices also, since the "Plr" voices share samples with their respective solo voices.

    There was a lengthy thread on this topic a year or so ago where an alternative was proposed that relied on detuning instruments a half step in either direction so that multiple instances of the same instrument could be used together. However, because the samples, even in the solos, are not always chromatic, the limitations I've described above still existed.

    The current arrangement of GPO makes it difficult to achieve divisi to the exent that you require. I'm hopeful that additional alternatives--1/2 sections and more solo instruments, for example--will be available in the next version of GPO that won't require the careful management of the instruments in order to avoid the above problems.
    Bill

  6. #6

    Re: Variable string section size / harmonics

    It's very gratifying to have received such good, meaty responses so soon after posting!

    I think I'd already come close to Nikolas's conclusion that my only recourse might be to use nothing but solo parts, but of course Bill rightly points out some of the problems with doing that. I'll have to experiment with various options: very slight and subtle detuning, use of the variable timbre and intonation controls (although I appreciate the latter is mainly intended for fast-moving passages), 'humanization' of timing and possibly other aspects such as velocity and so on - perhaps even experimenting with that idea of detuning by half a step and transposing, which hadn't occurred to me as a possible answer.

    Using the 'plr' instruments sounds like a good idea, Bill, but I'm rather relying on the KS instruments at the moment - and indeed there seems to be no other way to achieve muted sounds (which are lovely, btw!). At least there are three each of these for solo violin and cello, but unfortunately only one viola and bass. I'll just have to see what I can do! And yes, I too am looking forward to finding out what's in the vaunted GPO Advanced, or at least the next version of GPO. At the very least I'd like some string harmonics and senza vib, the latter of which I simply have to abandon at the moment.

    And re harmonics, Nikolas, I too assumed pitch shift would be a no-no by octave amounts, but when I tried it (albeit briefly) I was surprised that it sounded a lot better than I expected. It would of course be terrible if shifting downwards, but upwards seems to be not too bad. The band pass idea is interesting, but that would be extremely hard to implement for harmonic glissandi, of which I made a big feature in this piece, as the harmonics produced during the gliss range over a twelfth, so the filtered fundamental frequency would have to keep moving - not easy! And these are some of the passages where the players are all doing it in their own time to produce a random texture, which only compounds the problem. (I have already recorded all the solo parts for these passages, playing the same arpeggio repeatedly - I really thought I was in for repetitive strain injury!) What I will try is this: I'll see if pitch shift can be made usable, by means of some EQ or even filtering to reduce not the bottom of the sound but, paradoxically, the top. This is entirely an instinctive idea, and it may not work, but I feel it would need some of the 'shine' taking off it to produce that feathery, whiffly sound.

    Anyway, I'll report back here when I've tried all these things.

    And no, sorry Nikolas, much as I'd love to let you see the score, it's unpublished, very large, and all I have is a couple of copies of my original handwritten work from thirty years ago. Another project on my horizon is to do a new score in Finale, at which point maybe the odd page might find its way onto my web site - but I have to redo the web site first, so don't hold your breath!

  7. #7

    Re: Variable string section size / harmonics

    Quote Originally Posted by billp
    If the samples are triggered exactly simultaneously on unisons, the apparent volume of the duplicated sample will simply double without creating a natural texture of two instruments. If not simultaneous, you will hear unnatural phase cancellation (I think you mentioned this). You will have this difficulty if you intermix the "Plr" voices with their respective solo voices also, since the "Plr" voices share samples with their respective solo voices.
    Please remember that this is only a problem in the rare cases where the two voices share the same pitch and at the same velocity layer. I find it easier to go ahead and use the same samples, and then to fix any of the very few cases individually.
    - Jamie Kowalski

    All Hands Music - Kowalski on the web
    The Ear Is Always Correct - Writings on composition

  8. #8

    Re: Variable string section size / harmonics

    Well, I said I'd report back, so here we are...

    I've now done all 24 of my violins as solo parts, and it works very well indeed. I've set them to the three different KS solo violins in turn, so of each batch of 12:

    1, 4, 7 and 10 are the Gagliano
    2, 5, 8 and 11 are the Stradivari
    3, 6, 9 and 12 are the Guarneri

    1 to 3 are tuned +2.0 with the parts transposed down a tone
    4 to 6 are tuned +1.0 with the parts transposed down a semitone
    7 to 9 are tuned normally
    10 to 12 are tuned -1.0 with the parts transposed up a semitone

    An extra advantage of this tuning is that there is an occasional top E flat in the upper parts, which is just out of range normally, but now of course, it plays. I'll be able to do something similar to this with the three solo cellos, but if I'm going to use KS, which I really need to, I'm stuck with just one each for the violas and basses, so I'll have to see how that goes.

    Where the players are playing the same part, I've used Digital Performer's 'humanize' function to move the notes a few ticks randomly backwards or forwards, which does give a more realistic effect anyway. Separating the parts out like this also gives me the chance to 'dirty up' pitch bends and portamenti a bit for added realism.

    One thing I'd forgotten was that I'd have to redo all the tremolandi: whereas for the section strings one sustained note will do it, I had to play every single one of the solo parts - oh joy! Worth it in the end though.

    As far as Bill's comment about what happens with simultaneously triggered samples, etc. is concerned, all I can say is that I've experienced absolutely nothing of the sort. I definitely haven't noticed any doubling of volume either (but then what, strictly speaking, does that mean?). Of course the volume increases as you add more players, just as it does in a real orchestra, but it certainly doesn't perceptually multiply by the number of players. I'm also not aware of any 'unnatural phase cancellation', which I would expect to sound pretty nasty.

    On the whole, I'd say doing it this way, while of course it's a heck of a lot more trouble, produces a better end result than using the section sounds. The exception is the muted sounds, which aren't as nice and lush as the section ones, which is a pity, but I don't use the mutes a huge amount in this piece, so I can live with that. It also seems that you can't do portamento with solo mutes, so where that goes beyond the range of pitch-bend I've had to return to the normal sustain sound and just reduce the volume a bit.

    An an extra little side benefit, this setup allows me to pan each part slightly differently, to correspond as far as possible to a real orchestral layout.

    Finally, harmonics: yes, this really is a problem that I haven't solved satisfactorily. I've done as I suggested above and used a combination of pitch shift and filtering, but it's far from ideal. The main problem is vibrato. It's pretty damn hard to play an artificial harmonic with vibrato, and completely impossible for a natural harmonic, so until we have a 'senza vib.' KS - or even better, one for harmonics - it's not going to be possible to do this at all realistically. Allowing for that, it's not too bad for single, sustained notes, but those pesky harmonic glissandi do end up sounding very strange! My intention of course is to create a sort of ethereal, other-worldly atmosphere, and to that extent I suppose it works. It just doesn't sound much like harmonic glissandi, or indeed like orchestral strings in any form! Also, by the time you get up to the 9th harmonic on the first string it is extremely high (F#7), and as I really didn't want to push pitch-shift beyond an octave, all I could do at that point was to tune the instruments +2.0 and apply maximum pitch bend.

    So that's it for now, and thanks to those who chimed in with their comments. On to the violas...

  9. #9

    Re: Variable string section size / harmonics

    I haven't tried this and am not sure how well it will work, but...

    ...for harmonics, how about mixing in some Flute Solo NV? At higher
    registers, the 'pure' flute tone might give the illusion of harmonics.

    - k

  10. #10

    Re: Variable string section size / harmonics

    Thanks for the harmonics idea, klassical - I might just give that a try. As I've discovered since I posted my summary (at which point I'd only done the violins), my 'technique' gets worse the lower you go, and by the time you get to the lower strings of the cellos it just produces horrible distortion! I think I might end up experimenting with a hardware synth and abandon GPO altogether for that purpose (as I already have for the one place where I need pizzicato glissandi over an octave on the cellos - I used the acoustic bass patch on my Korg 03R/W with an added falling off in volume on each note, and by the time it's in the mix with everything else, I think it'll be OK).

    Another unforeseen problem has cropped up. Obvious, of course, but I just hadn't thought of it: where I've detuned some of the instruments sharp and transposed the parts down by the same amount, if the part goes down to the very bottom of the instrument, not only is the lowest note or two not sounding, but on the viola and cello I'm getting some pretty bizarre effects by introducing playable trills with the keyswitches on B and B flat!

    To combat this it would be nice if I could change the tuning on the fly, but the standard MIDI controller combination assigned to tuning (RPN [101 and 100] set to 0 and 1, data entry [6] to 64 +/- however many semitones) appears not to be implemented. Am I correct in thinking there's no way of changing the tuning on the fly? Otherwise, given that I'm committing this to audio in short chunks anyway, I'll just have to change the tuning manually as I do so, and keep notes of where I've done it.

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
  •