Approching Strings

[Frederic Petitpas]

So I had this project with like a violin/viola/cello/ensemble1/ensemble2/ with like pizz/stac/short and long bow patches and maybe another one with a "full ensemble patch".

The main thing consisted of chords and a melody so I took everything and combined them into a single track, then cut the melody out and pasted it in it's own track and did the same with the pizzocato parts.

I end up with 3 tracks (will maybe have 4) :

Chords (Full Sections patches)

Melody (23 violins patch) always made with single notes, no harmonies/chords in there

Pizz (pizz patch)

The fourth will probably be a Short Bow patch of some sort.

Now I need to know if worker this way will work with something like EWQL or do I really have to go anal about it and have a patch for each instruments in the "chords" track with every articulation on a separate track or do you really just go with a good patch and keyswitches if needed ? I saved a lot of CPU power which I really needed

Anyway, searched the other threads regarding strings and people were focusing on humanization and whatnot but that's not my concern here. It's all composed in an editor but I intent to go back and play the part on my midi piano once evertying is clean/mixed.

Edited April 16, 2014 by Metal Man

[ShrackAttack]

If you're not worried about humanization or it sounding real, then yes you can just load up a few EWQL patches and plop the MIDI in.

Edit: Regardless, things tend to sound better when individual lines have their own patches. The full section EWQL patches don't exactly sound great.

[Meteo Xavier]

The full section EWQL patches don't exactly sound great.

I disagree, they sound fine for the right job, they're just not for everything.

[timaeus222]

Edit: Regardless, things tend to sound better when individual lines have their own patches. The full section EWQL patches don't exactly sound great.

I can see what you mean. The full section patches are already set to sound like a limited set of variations in timbre, but combining a bunch of separate related patches could even triple the variations in timbre, making it more realistically emulating the interactions of each instrument's sound waves (similar to sympathetic resonance on vibrating bodies of string instruments).

[Frederic Petitpas]

That's the approach I wanted to have at first but I don't compose instrument per instrument when it comes to strings. I see them like chords on a guitar with an optional additional melody (solo violins). I'm more and more leaning toward the use a midi keyboard, I love the unquantized effect and control and all, but those old files are chords from Guitar Pro where I had like drums/guits/bass/keys and then maybe a patch of Strings Ens1 and another one of Strings Ens 2 + a violin. I decided to combine Strings Ens1 and 2 because they were essentially playing the same thing minus a few differance.

So would you have like 25 violins, 25 violas, 25 cellos, 25 contrabasses for a total of 100 tracks for chords ? (I'm exagerating the numbers here.. I didn't check to see how many of each is needed). But anyway, my computer would die even in the afterlife with that many tracks

Edited April 16, 2014 by Metal Man

[Pl511]

Because I work on a laptop, I have to deal with problems of CPU and RAM usage pretty often, and it's changed the way I deal with strings a lot. I also use a MIDI keyboard for recording input and then adjust it manually by hand afterwards. Because of this, I now see my strings library as essentially 8 parts - One violin, one patch with multiple violins; one viola, one patch with multiple violas, etc. I also mix down to .wav everything after I complete it, and move the original midi recording to a separate muted instrument track titled something like 'Violin Arco Melody' or 'Cellos Staccato' and so on. This means that I've essentially come to think of each part of the strings as a separate sequence of chords (and perhaps a melody) that could be played using just two hands (or perhaps 3 if I've manually added in additional notes) on a keyboard. While this doesn't necessarily give me a huge deal of room to work with in some instances, I've got a sheet of the ranges of each individual patch for each part of the ensemble, and I can use this to exploit each instrument as fully as I can to create a full sounding texture and harmonies that can be written independently to the melody.

Regardless of how workaround-y this system actually is, I find that usually some of the best writing for strings comes from when you write parts independently and then add them together. If anything doesn't fit, you have room to change it. If something you hadn't considered might work well fits nicely, you've just created something awesome at only a very small cost of a couple of seconds.

[timaeus222]

That's the approach I wanted to have at first but I don't compose instrument per instrument when it comes to strings. I see them like chords on a guitar with an optional additional melody (solo violins). I'm more and more leaning toward the use a midi keyboard, I love the unquantized effect and control and all, but those old files are chords from Guitar Pro where I had like drums/guits/bass/keys and then maybe a patch of Strings Ens1 and another one of Strings Ens 2 + a violin. I decided to combine Strings Ens1 and 2 because they were essentially playing the same thing minus a few differance.

So would you have like 25 violins, 25 violas, 25 cellos, 25 contrabasses for a total of 100 tracks for chords ? (I'm exagerating the numbers here.. I didn't check to see how many of each is needed). But anyway, my computer would die even in the afterlife with that many tracks

It would be quite RAM-intensive to just have a new instrument instance per note. If you can, sure? It'd be a lot of work. I would actually just rhythmically offset each note in the chord, write corresponding velocities depending on the pitch of the note, and layer on articulations. So, for example, perhaps I'd layer contrabass harmonics with contrabass legato sustains since the harmonics can go as low as 40~60Hz and the sustains can be high passed above that range and still sound strong at around 60~200Hz or so, forming a strong sustaining bass presence (you can route to a single bus and EQ before you get to the bus. Signal chains! ).

Sidenote regarding my own laziness: Although I wouldn't always recommend the following, sometimes I link many expression controllers (CC11 in EWQL and many other orchestral libraries) of similar articulations of similar instruments to the same automation clip so they're in sync; that way the swells are quickly editable and not wildly different. The rhythmic offset of the notes would offset the phase of the automation clip usage, but the automation itself would still be consistent. The reason I would recommend otherwise is just a small thing; for more realism, you'd be doing separate automations for separate instrument expression controllers because we're not synchronized robots, but I suppose if the difference in the results seems small to you, either way could be fine, but it's up to you and the context. i.e. if the strings are not exposed, it's less noticeable. It's like manipulating statistical data to make reasonable approximations and simplifying your calculations.

As for writing with a MIDI keyboard, it really depends on what one you have and how much memory you have. Its velocity response and your computer's response (i.e. delay between your note and the recording of the note on-screen) get rather interrelated. I usually just sequence strings, but if you have the memory, yeah, a good MIDI keyboard is worth getting to use for this.

Edited April 16, 2014 by timaeus222

[Moseph]

That's the approach I wanted to have at first but I don't compose instrument per instrument when it comes to strings. I see them like chords on a guitar with an optional additional melody (solo violins). I'm more and more leaning toward the use a midi keyboard, I love the unquantized effect and control and all, but those old files are chords from Guitar Pro where I had like drums/guits/bass/keys and then maybe a patch of Strings Ens1 and another one of Strings Ens 2 + a violin. I decided to combine Strings Ens1 and 2 because they were essentially playing the same thing minus a few differance.

So would you have like 25 violins, 25 violas, 25 cellos, 25 contrabasses for a total of 100 tracks for chords ? (I'm exagerating the numbers here.. I didn't check to see how many of each is needed). But anyway, my computer would die even in the afterlife with that many tracks

The short answer is to do whatever gets the job done with results you like.

If you want to be complicated with it, then you should do whatever allows you to access all the articulations you need for each of the five orchestral strings sections (violins 1, violins 2, violas, cellos, basses). So yes, either keyswitches for multiple articulations on each section's track or multiple tracks per section with one articulation each. Each way has benefits and drawbacks. I think EWQLSO is set up more to work with one articulation per track, but I know it has at least a few keyswitched sets.

It's a pain, especially with larger libraries. My EWQL Hollywood orchestral template was so large that I ended up splitting it into one project file per instrument/section and then just bouncing things down to import into a master project with just audio and no VSTs. SONAR really didn't like my trying to fit everything in one template (something like 2000 tracks including the many, many audio tracks for mic positions fed separately out of many instances of PLAY). (EDIT: I wouldn't have been running everything at once because I don't have the computer for it -- I'd have been loading samples section by section then bouncing down like I do now, but just in a single large project.)

It's enough of an issue that even in the pro market, there are companies offering libraries that combine all of the string sections into one big blended section to make it easier to write quickly without getting bogged down in library minutiae.

Edited April 16, 2014 by Moseph

[ShrackAttack]

I disagree, they sound fine for the right job, they're just not for everything.

Well yeah, it works fine for some things. But when you play these huge ensemble patches, you may have individual lines in your chords going to different instruments because of how the patch is laid out. Because you have something playing technically in the "cello" range of that patch and you move up to the viola range, it may move that line to now be panned towards the direction of the violas (since you're now in the viola register of the patch), etc.

For example, the 50 Piece str Sec Sus patch in EWQL gold pops the upper strings right on a G just randomly, so if you're playing in that register it is going to sound like shit, because you'll have instruments just popping in and out. So you just have to plan out where you're going to be playing so it doesn't sound ridiculous.

So would you have like 25 violins, 25 violas, 25 cellos, 25 contrabasses for a total of 100 tracks for chords ? (I'm exagerating the numbers here.. I didn't check to see how many of each is needed). But anyway, my computer would die even in the afterlife with that many tracks

A lot of people will have tracks dedicated to Violin 1s, Violin 2s, Violas, Cellos, DB for different articulations that you plan on using. A simple way to do is to just do Sustains and Stacs/Spics and add anything that you may need later (pizz, sordino, tighter spics, trems, etc). Honestly, the only important thing is the end result, and if you spend enough time you can make most anything sound good. EWQL should be fine for whatever. For example, most RPG music just has a low string patch and a high string patch and doesn't pay much attention to individual sections of the string section anyways. Obviously it isn't going to have that completely real film score vibe going on, but it's got that different RPG vibe which sounds cool .

Honestly, whatever you do, just make sure you give it some life. Use some expression and mod wheel to make it breathe.

Edited April 16, 2014 by ShrackAttack

[Frederic Petitpas]

You guys amaze and inspire me

Given that all I'll do is one little prologue with strings and the rest will just support a metal album, I'll arrange stuff real basic.

I don't think it calls for a very symphonickess approach but I'll still have to take care of what I'll do

You can hear the strings before the merging in the first 5 minutes of this video if you're curious, but I don't have the updated version discussed in the OP yet :

www.youtube.com/watch?v=z-58r4GXIXM

Note that this isn't humanized and anything but it'll show the kind of tune/theme involved.

I'll be using Miroslav. EWQL will be added to my tools one day but probably not this year, so I'll have to either do it with Miro or send it to someone once eveything will be prepared (arrangement/humanisation/mix).

Edited April 19, 2014 by Metal Man

[mickomoo]

In EWQL I've mostly stuck to assigning an articulation per track. To "properly" emulate the orchestra I use patches that have generally the same number of instruments (per section) you'd hear in a symphony, which I think are among EWQL section patches.

[Neifion]

Late to the party here, but if you're wanting to go for a realistic sound, you really should write the way composers do for live orchestra. I know the OP said he was just doing basic stuff to support the metal work, and so obviously I'm not suggesting he should go all hyper-orchestral and all. Of course this is more to offer suggestions to those who are going for the orchestral concert sound, whether it be cinematic or classical.

First, you should write for each section: V1, V2, Vla, VC, and CB separately. Play each in separately, and control CC dynamics separately, one at a time, by hand as well. And no quantize. All this goes a long way for humanization. Use proper section sizes (standard symphonic size is 16, 14, 12, 10, 8, standard chamber size is 4, 3, 3, 3, 3, though the number can vary slightly from library to library). And also know what string sections are used for what. For instance, V2 are typically used for harmony, not for melody. Stay away from ensemble patches, as in a real situation, you will virtually never hear all 5 sections playing the same note. At most, 2 or 3 sections may converge in a passage, perhaps octaved or not, to highlight a motif, chorus, or during an important crescendo, but never for long, because it gets rather static fast . Orchestral compositions are dynamic, highlighting certain sections at times, allowing two or three to harmonize for a moment, before each breaking off to do it's own thing, then converging for an important motif, etc.

Of course, as people have said, having one track per articulation, multiplied by 5 sections, can be a lot. That's why I do similar to what Shrack mentioned: I have a template with just the legato, stacc, and pizz, so 15 tracks. Then, if I need additional ones like a Vla trill, or V1 flautando, I'll add them in as needed. If you're on Cubase, you can route multiple MIDI tracks to a single VST instrument, with each MIDI track routed through it's own MIDI channel into the separate articulation patch (legato, spicc, etc.) Not sure how other DAWs do it, but it makes life a lot easier, especially when I'm dealing with the rest of the full orchestra. Some people are fans of the keyswitch approach; that I am not.

Another benefit of writing and playing in each section separately is that it teaches you the unique qualities that each section has to offer. Viola, Cellos, CBs all say things differently; they each provide different colors and emotions. You can slap them all on at once, sure, but if you really want to maximize the quality of your strings, you need to learn how to paint each one separately. That way you can do things like write in a cello line to emphasize a bit of dramatic depth, or a pp-mp violin trill here to add a delicate, wintry feeling, or a smooth viola legato sound like floating through the air.

Also, in talking to some of the guys at EW and Spitfire, the ensemble patches are mostly there for quick idea sketching, with the intent that you'd go back and use the individual section patches to get the final sound.

One thing to remember as well, is that if you have a V1 patch loaded, for example, when you press a key, that's all violins in that section playing at once. If you want to do a chord, simply split the notes of the chord so that each section only plays one note. For example, a three note chord can have the violins playing the high note, the violas playing the mid note, and the cellos playing the low note. If you need to have one string section playing more than one note of a chord, the best thing would be to use a divisi patch (section size of 4-8 for example). I admit, sometimes I fudge a little bit, you go by your ear of course, but if you're going for realism, rule of thumb is to be aware of those sections sizes and not go over the max amount too much, else the ear will subconciously start to notice. And I wouldn't even think about using an ensemble patch for chords!

And then there's panning. From left to right: V1, V2, Vla, VC, CB. High to low. If you're hearing more violin in the right ear than the left, something is wrong.

Finally, the ear is king of course, so if it sounds good, screw rules and go with it. But if you're doing an orchestral piece and your goal is to get a convincing orchestral sound, you definitely want to consider the rules. I tend to stick to the rules even when I do hybrid tracks, but that's a style thing.

Some examples:

Classical

https://soundcloud.com/kekopro/aires-and-graces

Cinematic

https://soundcloud.com/kekopro/the-life-he-gave

Hybrid:

https://soundcloud.com/kekopro/blind-ambition

Hope this helps!

Edited April 25, 2014 by Neifion

[timaeus222]

If you're on Cubase, you can route multiple MIDI tracks to a single VST instrument, with each MIDI track routed through it's own MIDI channel into the separate articulation patch (legato, spicc, etc.) Not sure how other DAWs do it, but it makes life a lot easier, especially when I'm dealing with the rest of the full orchestra. Some people are fans of the keyswitch approach; that I am not.

It's literally done conceptually the same way in FL Studio, so there's that confirmation. You can set a VST to take a particular MIDI I/O and all the MIDI channel instances can each be routed to individual mixer tracks, each channel allowing MIDI CC like expected. And yeah, I also think that keyswitches just add another factor to doing it conventionally.

Another benefit of writing and playing in each section separately is that it teaches you the unique qualities that each section has to offer. Viola, Cellos, CBs all say things differently; they each provide different colors and emotions. You can slap them all on at once, sure, but if you really want to maximize the quality of your strings, you need to learn how to paint each one separately. That way you can do things like write in a cello line to emphasize a bit of dramatic depth, or a pp-mp violin trill here to add a delicate, wintry feeling, or a smooth viola legato sound like floating through the air.

Yeah. I still haven't actually done a truly full orchestral track yet, but I've been focusing much more on imagining each instrument being played and considering the possibility of highlighting particular parts at times in a piece. I think that makes total sense.

One thing to remember as well, is that if you have a V1 patch loaded, for example, when you press a key, that's all violins in that [patch] playing at once. If you want to do a chord, simply split the notes of the chord so that each section only plays one note. For example, a three note chord can have the violins playing the high note, the violas playing the mid note, and the cellos playing the low note. If you need to have one string section playing more than one note of a chord, the best thing would be to use a divisi patch (section size of 4-8 for example). I admit, sometimes I fudge a little bit, you go by your ear of course, but if you're going for realism, rule of thumb is to be aware of those sections sizes and not go over the max amount too much, else the ear will subconciously start to notice. And I wouldn't even think about using an ensemble patch for chords!

It's a subtle difference in polyphony, but most of the time it's a good idea. However, I think it's fine to do it the 'easy' way with ensemble patch + chords if the strings are not very exposed. That's just an excuse though, and not really the convention. It does save you some time though, if you're feeling like you're short on time for something. Kinda depends on the library quality, but yeah.

Finally, the ear is king of course, so if it sounds good, screw rules and go with it. But if you're doing a fully orchestral piece and your goal is to get a convincing orchestral sound, you definitely want to consider the rules.

And that would be my approach. Edited April 25, 2014 by timaeus222

[Frederic Petitpas]

I'm not sure if I understand what you mean with the "V1" thing and your 15 tracks for articulation. lol

You have V1, which contains 16 violins, and use 15 articulations for each ?

16 * 15 = 240 tracks for the violins.

So, with all the sections : 16, 14, 12, 10, 8, each * 15 = (16*15)+(14*15)+(12*15)+(10*15)+(8*15) = 240+210+180+150+120 = 900 tracks

I do not think that was what you meant. Would that be 15 patches ?

Let's separate the instruments from the articulation one moment :

You need 16 violins, do you load 16 patches ? Because I figure you probably want to pan them with a little diferance.

[Neifion]

You do not have 1 patch per player. You have 1 patch per string section (i.e. 1 patch for V1, 1 patch for V2, etc.)

You have 5 different string sections. V1, V2, Vla, VC, CB.

You have one legato patch for each section. That's 5 patches.

You have one staccato patch for each section. That's 5 patches.

You have one pizzicato patch for each section. That's 5 patches.

Total of 15 patches, each in it's own track in the DAW.

Of course, only one articulation per section can be playing at any time. So at the maximum, you'll hear V1s playing one of the three articulations (legato, staccato, OR pizz), the V2s playing one of the three articulations, the Vlas playing one of the three articulations, and so forth. So at most, each of the 5 string sections will only be playing only one of their three available articulations at any given moment. The reason I have all three articulations up in the template is so that I can switch articulations whenever I need to.

For instance, if the V1s are currently playing legato, I won't have them also playing staccato or pizzicato. That way, only 16 violins maximum are playing at one time, just as in a real orchestra. And no matter what articulation they are currently playing, they can only play one note at a time (no chords). If there is a chord in the composition, it is composed of multiple string sections, each section playing one note of the chord (as I described in my last post). The exception is that if I do want a single section to play a chord, I use divisi patches (smaller number of players) to maintain the overall section size.

For instance, from 1:08 - 1:28 of the following song, I have V1 playing melody legato (1 articulation/1 patch), Vla playing counterpoint legato (1 articulation/1 patch), VC playing staccato (1 articulation/1 patch), and CB playing pizzicato (1 articulation/1 patch). If I wanted to add anything else, the only thing I could add would be V2 (choice of legato, staccato, OR pizz), but I decided to let them take a break for that segment. So only 4 patches are playing at once, 1 patch from each section (except for the V2s).

And then from 1:28 - End, the V1s switch to playing pizzicato, the Vlas also switch to playing pizzicato, the VC switch to playing legato, and the CB continue playing pizzicato. So 4 patches are still playing at once, but 3 have switched to a different articulation patch.

https://soundcloud.com/kekopro/aires-and-graces

Hope that made sense!

I'm not sure if I understand what you mean with the "V1" thing and your 15 tracks for articulation. lol

You have V1, which contains 16 violins, and use 15 articulations for each ?

16 * 15 = 240 tracks for the violins.

So, with all the sections : 16, 14, 12, 10, 8, each * 15 = (16*15)+(14*15)+(12*15)+(10*15)+(8*15) = 240+210+180+150+120 = 900 tracks

I do not think that was what you meant. Would that be 15 patches ?

Let's separate the instruments from the articulation one moment :

You need 16 violins, do you load 16 patches ? Because I figure you probably want to pan them with a little diferance.

Edited April 26, 2014 by Neifion

[Frederic Petitpas]

Hey thanks for the explaination !

One note per section to form chords with other sections if something I've overlooked. I did think about it, but I often just copy/pasted/transposed in octave chords to other sections. I did it with single notes and separated sections to form chords but I did not know that it was a standard (most of the time). It's very good to know !

I'm really curious about the divisi patches. As I said, I only have Miroslav for now.

Finally, must I understand that you record a performance for each section and that for 16 violins, you have a single performance but a 16-violin patch ?

[Neifion]

Yep, you got it. My V1 patch is a single patch containing the entire 1st Violins section. Same idea for the other 4 sections.

Hey thanks for the explaination !

One note per section to form chords with other sections if something I've overlooked. I did think about it, but I often just copy/pasted/transposed in octave chords to other sections. I did it with single notes and separated sections to form chords but I did not know that it was a standard (most of the time). It's very good to know !

I'm really curious about the divisi patches. As I said, I only have Miroslav for now.

Finally, must I understand that you record a performance for each section and that for 16 violins, you have a single performance but a 16-violin patch ?

[Moseph]

I'm really curious about the divisi patches. As I said, I only have Miroslav for now.

Divisi patches only exist in some of the high-end libraries. Having them basically means doubling or tripling the sample content, since whoever makes the library has to record everything multiple times with the different numbers of players. The libraries I can think of off the top of my head that I know have divisi are LA Scoring Strings (Full), Hollywood Strings (Diamond), and Adagio Strings. I have actually had a bit of success using Session Strings Pro as a makeshift divisi ensemble for VSL strings, since Session Strings Pro has fewer players per section than VSL, but it's harder to get a proper blend doing this than with an actual divisi library.

Edited April 26, 2014 by Moseph

[Neifion]

I use Spitfire Sable as my divisi (chamber-sized sections of 4,3,3,3,3). Sable is a library of it's own though; doesn't come with the full symphonic sections (for symphonic-sized I use Spitfire Mural or Cinematic Strings 2). My wallet is still limping almost a year later, but since I mostly do involved orchestral compositions, it was worth it as they sound amazing. I often actually prefer the chamber sound to the symphonic sound in my writing, so Sable is my go-to for strings these days.

I believe LASS has auto-divisi, which will actually split the players automatically if you play a chord (correct me if I'm wrong?) Don't have LASS, so I'm just going off what I remember hearing.

Divisi patches only exist in some of the high-end libraries. Having them basically means doubling or tripling the sample content, since whoever makes the library has to record everything multiple times with the different numbers of players. The libraries I can think of off the top of my head that I know have divisi are LA Scoring Strings (Full), Hollywood Strings (Diamond), and Adagio Strings. I have actually had a bit of success using Session Strings Pro as a makeshift divisi ensemble for VSL strings, since Session Strings Pro has fewer players per section than VSL, but it's harder to get a proper blend doing this than with an actual divisi library.

Edited April 26, 2014 by Neifion

[timaeus222]

I believe LASS has auto-divisi, which will actually split the players automatically if you play a chord (correct me if I'm wrong?) Don't have LASS, so I'm just going off what I remember hearing.

https://www.youtube.com/watch?v=CI_a7cnmoRM

[Neifion]

Yep, there it is.

https://www.youtube.com/watch?v=CI_a7cnmoRM

[Frederic Petitpas]

So I made a template with Miroslav.. For the violins all I had was either 23 or 11 so I took both as the panning isn't exactly the same. Then Violas, cellos, basses. I left them panned as it is already in Miroslav and have 8 variations in my template (full, stacc, pizz, legato, etc..) for each.

I did something quick with the keyboard, it's just one variation per section only but to test the panning and the sound.

https://db.tt/8FAPvZBW

Now, what's the relation between the variations and the sections in general ? In the clip above I have something like the 23 violins playing trem, 11 violins stacc, violas pizz, cello short bow, basses long bow

Template looks like this :

Only played 5 minutes with it so far but I can see how pleasing it is.. will complete the template with other orchestral elements.

Edited May 2, 2014 by Metal Man