# Topic: Divided strings with GPO4

1. ## Divided strings with GPO4

I've posted a document called "Creating Divided String Sections of Arbitrary Size with Garritan Personal Orchestra" to my web site. You can find it here:

http://www.johnmelcher.net/Assets/PD...O4 rev2012.pdf

I'm very interested to hear comments, corrections, etc.

John Melcher

2. ## Re: Divided strings with GPO4

quick off the top.. the first Strawinski excerpt has some errors to your division of the players.

the violas are divided into section + 1 desk (not equal parts 6+6 as you have).

the cello section is likewise divided 2 desks divisi, then "gli altri" (the others).

you can't really put absolute numbers on the divisions. for the celli, for example, going by your standard numebr, it would be 2 desks (4 players) on a divided part, and the balance (6 players) on the other part. It can never be "5+3+2" as you write. It must, by its very definition, be 2+2 +(however many more/less make up the section).

3. ## Re: Divided strings with GPO4

Originally Posted by qccowboy
quick off the top.. the first Strawinski excerpt has some errors to your division of the players.

the violas are divided into section + 1 desk (not equal parts 6+6 as you have).

the cello section is likewise divided 2 desks divisi, then "gli altri" (the others).

you can't really put absolute numbers on the divisions. for the celli, for example, going by your standard numebr, it would be 2 desks (4 players) on a divided part, and the balance (6 players) on the other part. It can never be "5+3+2" as you write. It must, by its very definition, be 2+2 +(however many more/less make up the section).
Thanks for these comments! Indeed, the violas and cellos are divided as you said. In my weak defense, I was reading off the Dover miniature score and completely missed the "1 P." and "1,2 P." markings.

It's certainly possible to divide desks to get, for example, 10 celli divided 4-3-3. Though I can't recall a composer specifying an exact split except for total divisi (Penderecki's "Threnody" or Strauss' "Metamorphosen").

4. ## Re: Divided strings with GPO4

John, your article is a great help to me. Thanks. A question: what do you personally use to get harmonics or non-vibrato. Also, what solo string set do you use? GPO4 is just not adequate by itself for my purposes.

John

5. ## Re: Divided strings with GPO4

bkkjohn,

Sorry to say, but your paper is VERY misleading, and offensive towards most of the competitive libraries mentioned. I`d suggest to remove all false statements, before you get in any trouble, or even better - all mentions of anything besides GPO, being the focus of your article.

Although, on personal note, the GPO parts are misleading as well, professionally-wise. BTW there were papers on this subject in the past - I bet you would do more good by sticking to those.

But as I said, false claims are offensive, specially cause they undermine the labor of everyone in the market. I think Garritan knows better than others, how those urban myths gather over the years.

6. ## Re: Divided strings with GPO4

bkkjohn,

Sorry to say, but your paper is VERY misleading, and offensive towards most of the competitive libraries mentioned. I`d suggest to remove all false statements, before you get in any trouble, or even better - all mentions of anything besides GPO, being the focus of your article.

But as I said, false claims are offensive, specially cause they undermine the labor of everyone in the market. I think Garritan knows better than others, how those urban myths gather over the years.

What "false statements" were you referring to? A primary reason to make the document available here is to correct any errors. I wish you had provided some examples. You imply there are many; perhaps you can e-mail me your list. I plan to incorporate these chnges in a new version. Whenver the libraries' web sites or manuals have not been clear, I've talked to the manufacturers directly whenever possible. The paper may not be 100.000% accurate or complete, but it's not intentionally misleading. I'm very interested to hear personal examples of solutions. If you know of additional capabilities of one or more string libraries that I've missed, please do let me know!

My personal goal is the virtual orchestration of my own work, using string ensembles of any desired size, and not writing to conform to a library. My paper focuses on solutions to this particular challenge. GPO would win hands-down if the sounds were better. So far, I haven't found any acceptable solution without several more computers, or to multitrack the 70 DVZ library many times.

The last time I had to deal with complex virtual solo strings in orchestral writing was in 2005, and I used Kirk Hunter Solo Strings with Gigastudio. The sounds are beautiful, but too espressivo for orchestral playing (but they were great for a string quartet recording). Mostly I needed non vibrato playing, which was a problem because there were no cello NV instruments, only open strings (Oddly, spread chromatically across the scale), which had the wrong timbre and needed EQ. In about a year, I'll need to record an orchestral work in progress, and realistic string ensembles of the correct size are crucial. Thus, I'm trying to decide now where to make a considerable financial investment.

BTW, I was suspicious of the numbers for the NY Phil, rechecked and found an error. According to their web site, their string section comprises 63 players (31-11-12-9). That makes them virtually the same size as Chicago and LA.

7. ## Re: Divided strings with GPO4

bkkjohn,

Sorry to say, but your paper is VERY misleading, and offensive towards most of the competitive libraries mentioned. I`d suggest to remove all false statements, before you get in any trouble, or even better - all mentions of anything besides GPO, being the focus of your article.

Although, on personal note, the GPO parts are misleading as well, professionally-wise. BTW there were papers on this subject in the past - I bet you would do more good by sticking to those.

But as I said, false claims are offensive, specially cause they undermine the labor of everyone in the market. I think Garritan knows better than others, how those urban myths gather over the years.

Those are strong words, Vlad. Please provide examples of any misleading or "offensive" comments (I'm not convinced you know what that word means). I think the paper is reasonably accurate as far as the other string sample libraries are concerned. I've used LASS and Hollywood Strings, and the divisi features of those libraries are as described in the paper.

And the techniques he uses to "derive" additional players from the same samples has been discussed with regards to other libraries and was a mainstay back when computer storage limited composers to just a handful of samples per project (Garritan happens to write some of these techniques into the library itself).

************

John, as far as comments on the paper, I think you discuss the issues around divisi reasonably well. There is one library (if you can really call it a library, it's more like a synth-library hybrid) that can handle divisi as well or better than any other and that's Synful Orchestra. Not to detract from GPO (which is the focus), but you may want to research this product for your paper as well.

You may also want to run your samples through polarity monitoring plugins to get a more objective picture of any phasing issues that could occur when using Solo and Player patches together.

Regards

8. ## Re: Divided strings with GPO4

I think, in order to write a paper with overall string market review, as it was made, in my opinion, entirely unnecessary, one should at least own half of those libraries. Bashing other people`s work without even experiencing it, is not so smart.

Anyway, just from top of my head: HS is NOT built from "half-choir" samples, whatever that is, Lass is not "one of the largest" libraries, and there are no 10 solo instruments in it.
VSL Dimension Strings IS about divisi and the phrase that it "was recorded later" is completely irrelevant - All those big VSL sets weren`t recorded in one go! Moreover such phrase as "violins contain only one sample set" - is quite absurd, when each VSL full library contains dozens of articulations and velocity layers. The formality of calling your recordings "2nd violins" is not a virtue per se, and since VSL sections got recorded evenly seated across the stereo field, and supposed to be mixed to reflect needed position later - you never know was it first or second group. Definitely you wouldn`t tell 14 pieces group from 12.

GPO never had its 2nd violins "derived" from recordings of the 1st, and if you wanted to explain that it does use chromatic solo samples to create "ensemble instruments" - you should be more clear.

Again - I see no point making overall "strings" review, if owning just GPO. Your article`s scope doesn`t call for any comparison in the first place.

9. ## Re: Divided strings with GPO4

I think, in order to write a paper with overall string market review... one should at least own half of those libraries. Bashing other people`s work without even experiencing it, is not so smart.

Anyway, just from top of my head: HS is NOT built from "half-choir" samples, whatever that is, Lass is not "one of the largest" libraries, and there are no 10 solo instruments in it.
VSL Dimension Strings IS about divisi and the phrase that it "was recorded later" is completely irrelevant - All those big VSL sets weren`t recorded in one go! Moreover such phrase as "violins contain only one sample set" - is quite absurd, when each VSL full library contains dozens of articulations and velocity layers. The formality of calling your recordings "2nd violins" is not a virtue per se, and since VSL sections got recorded evenly seated across the stereo field, and supposed to be mixed to reflect needed position later - you never know was it first or second group. Definitely you wouldn`t tell 14 pieces group from 12.

GPO never had its 2nd violins "derived" from recordings of the 1st, and if you wanted to explain that it does use chromatic solo samples to create "ensemble instruments" - you should be more clear.
...

Vlad, You seemed to have misread and misinterpreted what I wrote. In fact, I'm not sure you even understand the problem.

FIRST: There are exactly TWO string libraries (that I know of) that can provide multiple string ensemble of any size - 14 violins divided into 7 desks of 2 players, for example, or 3 violas. GPO is one; the other is 70 DVZ Strings, but it allows a total of five ensembles. In the first Stravisnky example I provided, the 2nd violins (approximately 14-16 players) are divided into seven parts, 2 players per part, playing double-stopped harmonics. While 70 DVZ can make a violin ensemble of 2 players, it has a limit of 5 channels. The auto-divisi function won't work as written. In the bar marked #87, the first 3 beats are played by only 4 desks, not the full choir.

This is a relatively simple example. For a composition where all the parts are more independent, it's impossible.

SECOND: A "half choir" is half of a full "choir", the term for all of the violins, violas, cellos or basses in a string section. Most modern string libraries now provide at least two half-choirs, which can be combined to form the full group. LASS provides full, half and two quarters.

THIRD: I am trying to emulate a real orchestra, and the 1st and 2nd violins sound different. There are more 1sts to give a thicker (not louder) sound, and the 2nd violins are positioned differently, not just further to the right, but rotated. The real problem with not having unique 1st and 2nd violins is combining them into a unison.

FOURTH: Hollywood Strings does indeed have half-choirs, though my point was that it offers nothing smaller. http://www..com/Hollywood-Strings

FIFTH: I did not say that GPO 2nd violins are derived from 1st violins. My discussion is limited to the solo string instruments. There are 3 individual violin instruments, and from each, 3 derived instruments.

SIXTH: I was referring to multiple samples of the same articulation in order to combine them in unison. I was not talking about multiple articulations.

SEVENTH: The VSL web site describes the Dimension Strings (which I haven't used) thus: "The 24 string players... were recorded in homogenous groups (8 violins, 6 violas, 6 cellos and 4 double basses), capturing all of the human interactions and slight imperfections of a live performance, but this time we’ve used individual microphones for each player, which allows you to access each voice individually."

If this actually means that I can play 8 contrapuntally independent solo violin parts, or 4 parts of 2 players each, with different sarticulations, and have it play properly, then all I need is 2 instantiations and I'm all set! However, as I interpret their documentation, these are ensembles but the balance can be adjusted, If you know this to be incorrect, and that it's inded possible to divide a Dimension String section arbitrarily with the same independence as GPO, I'd appreciate hearing in detail how it's done. So far I haven't gotten a reply from VSL technical support about it.

EIGHTH: Indeed, LASS has only 5 solo instruments, not 10. That was a typo (the table is correct). At 24 GB, it is one of the largest string libraries.

NINTH: The recording techniques used in the different VSL libraries "may not blend properly." I don't know for sure because I haven't tried it.

I hope this clarifies most of your issues. As I said, if you have ideas for alternate ways of achieving the same tasks, please let me now.

10. ## Re: Divided strings with GPO4

Yes, bkkjhon, I know you meant no harm. But lets say - in such market sensitive issues, one better make reading comprehension as simple as possible, no? I am in no way language specialist, but I could be a good example of what average non-native reader gets from your article.

"For example, Hollywood Strings Diamond Edition (East-West/Quantum Leap, MSRP \$995) provides a 57-player string section with half-choir samples only, which are combined to get a full choir."

Not only I would suggest to avoid listing prices, as it is very dynamic and sensitive issue, what I read from here, is that MAIN Hollywood Strings sound is achieved by "combining" divisi samples - which is not. Would you put it that way, that in a Diamond edition of the lib, there is an optional patches type, which consists of an additional "divisi" mic, which prefers one side of the section over the other - I would react differently.

" LA Scoring Strings (Audiobro, MSRP \$1,399) is one of the largest (24 GB) ..."

EWQLSO strings ~ 30 GB
HS - 48-320 GB
VSL - Each full strings library ~ 30 GB
8Dio Adagio - Violins ONLY ~ 30 GB
Dimension Strings - Violins ONLY - 33 GB

Now about the Dimension strings - yes, there is personal control of each musician! You get 24 audio channels of studio string players. In my opinion - you better be a mixing pro to make this thing really shine.

" According to the LASS manual, 2nd violins are “derived From Violins I” and “programmed to avoid phasing issues”. This provides less depth and variety that separate samples, but it’s an acceptable
compromise, also used in GPO."

Why would I understand you meant solos? I did actually, cause I know all the ins and outs of both, but its the last thing GPO needs - to be misinterpreted about its 2nd Violins nature, which is in original samples of course.

Vsl again:

" Even at this size (111.4 GB) and cost (MSRP E4,280, ~\$5,500), there are issues. There’s only one set
of violin samples."

No - you can`t just say "one set". You have to understand one thing - VSL is not about venue recording as in Miroslav, or EWQLSO, or HS. VSL is about type of instrument/section in a studio project, so when you get a library, you get a whole bunch of stuff this section is supposed to do - Slured legato, detache legato , bow changes etc. - Those are all different sustains! So if you work with 14 viloins section - thats what it is, a violin section, if you want to make it 1st, 2nd, or award it a medal of a French Legion - its your own call. Just don`t say you miss alternative samples, when there are dozens of, with round robin and velocity layers included.

I understand your goals. Indeed, Lass, Dimension strings, DVZ - were specifically created to answer this need. Lets just be more focused, as there is much more to it, and give credit where due. Just by saying A>B you take responsibility of being credible source for your readers.