Strings sequencing/sampling/mockup [update]

[Cheez]

Those who are into orchestral mockups will know that there's no such thing as having enough string libraries. Each library has it's own strengths and weaknesses that result in us layering different libraries at times. Now, more and more string samples are appearing (esp this year and probably 2010). With so many libraries popping up, I thought it's time to review what's been happening in the area of strings.

First, let's lay some foundation. Some important factors to a good library include:


1. Realism. This is always the main point of any sample library. But to achieve realism is not easy. There are few problems inherent to string samples that need to be overcome:

• One articulation plays all. In the old days (and in most synths), it's still one string patch playing all articulations. This has changed tremendously. We have now almost all articulations (pizzicato, marcato, detache, legato, staccato, tremelo, trills, spiccato etc). There's also something called "true legato" of which the interval between 2 notes when played legato is actually sampled.
• "Machine-gun-effect". When play a note repeatedly, it sounds like a machine gun because the note triggered is the exact same note sampled. Solved by round robin - multiple samples of 1 note is sampled (usually one note sampled 4 times played at different times); when a note is triggered repeatedly, the 4 different samples rotate randomly so that each time, a different version of the same note is triggered.
• "Sucking effect". Because most string samples start with a slow increase in volume and decays the same way, when notes are played in a phrase, the tail-end of a note (release) and the beginning of the following note create a kind of "sucking effect" - very unnatural string playing. Almost all hardware synths has this problem. In samples, the best way to overcome this is via modelling (first achieved by Garritan in his Stradivari violin and Gofriller Cello). Clever programming (eg scripting within Kontakt) can also fix that.

3. Control (triggering articulations). Having so many articulations also mean that we need 10-20 midi tracks for strings alone when we need different articulations. That is tremendously reduced with key-switching (lower notes of the keyboard - each note will switch the entire keyboard articulation; eg, C1=legato, C#1=staccato, D1=spiccato etc) and modwheel crossfading (mod wheel in different position switches the articulation - eg position at lowest = legato; as wheel slowly moves halfway point, it crossfades slowly to tremelo etc).

3. Mic positions. Adding reverb sometimes just doesn't cut it. The delay tail is extremely important in a large hall. What's not there will never be there even with good reverb. Most libraries now have recordings with mics in different positions (from close to far).

There are a number of current great libraries out there that covers all/most of these aspects (EWQLSO, Sonivox, Kirk Hunter, VSL etc). I won't mention the outdated libraries here as these libraries superceded most of them in terms of all the factors mentioned above.


Now...drum roll....we have a new series of new libraries that came out/coming out. These include even more control and articulations that are never done before. One of the new hurdles that need to be overcome:

• Divisi playing. E.g. 1st violin need to play divisi (ie 2 notes). In a real orchestra, half of the 1st violins will play one note while the other half plays the other note. When there's no divisi, all the players play in unison. In sampling, it's not easily achieved automatically. If you are using a sample of 16 1st violins, for example, playing divisi of 2 means you get 32 1st violins suddenly. Playing divisi of 3 means 48 players - you get the idea. What we do currently is to use different samples with different number of playes - which also mean multiple midi tracks. New libraries try to overcome this by sampling different number of players per stand then trigger divisi automatically depending on how many notes are played at the same time.

The newer libraries are:

1. LA Scoring Strings (aka LASS). http://www.audiobro.com/html/lass.html
Not cheap. They recommend 2 PCs to run it well. The string "air sound" is simply beautiful in this library.
2. Cinematic Strings. http://www.cinematicstrings.com/
No auto-divisi, but it boast to be able to control samples easily and so eliminates the use of multiple midi tracks for one string line.
3. Kirk Hunter TVEC II. http://www.kirkhunterstudios.com/
Still waiting. Update from TVEC I programming. Divisi is added + new samples.
4. Coming out 2010 - Hollywood Strings by East West. http://www.soundsonline.com/content/...od-strings.php
New articulations galore (including bow change legato, richochet etc). Also not cheap.
5. Much anticipated and waiting...Garritan Orchestral Strings 2 (aka GOS2). Since GOS1 was taken out, people have been waiting for Garritan to release v2. Both the Stradivari and Gofriller were also taken out in anticipation of GOS2.

And of course, the tools are only a quarter of the battle to realistic mockups. Good arrangements and sequencing skills are more important...

[kongwee]

For keyboard player to learn about string layering, do not play string like playing piano. Basically, string section make up of Violin 1, Violin 2, Viola, Cello, Bass. Usually, you need to write each section at a time. A good arrangement do not use more than 4 notes at any time in the whole string section at a time. That exclude counter melody,"solo" and fill ins.

A string section use a wide range of note, it is important to plan them. Bass normally I like them to play roots note. Cello is not good at root when other Bass instrument is playing root. You have to share other 2 or 3 note of a chord among cello,viola and violin. As mention, write each section at a time.

Do check up for each string instrument what range do they play too.
Attend live string performances are very important to know and feel the articulation of real string player.
Lastly,your understanding of harmony is very important too.

[Cheez]

Ditto on harmony. I write the entire string section before sequencing (ie 1st/2nd Violins, Violas, Cellos and Double Basses). Each section influence how the other sections move.

Most sample library eliminates the problem of note range. They record only the range of the instrument, so if you play beyond the range, you get no sound...

[Cheez]

Oh, forgot. Just as important as harmony is counterpoint. And the combination of the two.

[KeDraiv]

Originally Posted by Cheez

Oh, forgot. Just as important as harmony is counterpoint. And the combination of the two.

hmms... so which should we start learning first?
harmony or counterpoint?

[kongwee]

Originally Posted by KeDraiv

hmms... so which should we start learning first?
harmony or counterpoint?

Harmony first!!!!.
I alway imagine myself with a group of string.
It easy for me as I perform with string before even I am not a string player.
Who should play and what articulation to use.
It is just harmonizing part is not so easy,as you add counterpoint,"solo" or fill in.

[Cheez]

Harmony. Counterpoint builds on it. Strong foundation in harmony is necessary.

[Cheez]

Ooh. Kongwee beat me to replying 1 min earlier!

[kongwee]

On midi side, you have long bowing on string. Do add expression automation or volume automation. A gradual up and down calve is need. It is more natural most of the time, unless you string sampler is that smart.

[Cheez]

Indeed. First of all, the library must have sampled both up and down bowing. And with the later libraries, if up and down bowings are included, they would have automated alternate bowing script to do it.