Wednesday, October 30, 2013

MIDI tutorials: string writing 101

One of the most enjoyable gigs I’ve had in recent years is arranging/writing “orchestral” backing tracks for my west-coast band, Helsott.  Much of my formal musical training (AKA “schooling,” which I find accounts really for little in the real world, but hey!) was as a composer.  However, I don’t think I’ve touched a piece of manuscript paper to compose in almost 15 years now! 

The fun thing about doing orchestral arrangements for a metal band, albeit sampled orchestral arrangements, is that it gives me a chance to flex those “classical” muscles I rarely use these days, and it brings that world right into collision with the world I mainly inhabit in my day to day gigging, that of music production.  Not only do I record and produce, but I also have worked as an instructor in vocational music production courses at the college level.  I find it really fun to share tips and tricks for production in the classroom environment; now, as the “mad metal professor,” I’ve decided to start blogging some of my favorites!  And here is my first installment...enjoy.

Create realistic sounding strings with MIDI sequencing:

1. Understand the ranges of the instruments that you’re trying to emulate!  A violin cannot play a D in the middle of the bass clef (D2 in MIDI language), and if your sampler doesn’t use this type of range constraint, you’ll need to KNOW what instruments can play what notes.  Get a textbook on instrumentation/orchestration; the two I recommend are Blatter’s “Instrumentation/Orchestration” and Kennan’s “Technique of Orchestration.”  These are both excellent tutorials on writing for strings and contain real-life examples of good string arranging.  One thing to be aware of, however, is that the designation for middle C in MIDI language and in “classical” writing is different.  In MIDI, C3 is middle C in most cases.  In classical training, it’s C4.  Don’t let that trip you up!

2. Understand typical voicings in orchestral writing.  In general, voicings in chords get tighter the higher the pitch.  Spread out chords.  Unless you’re going for a specific effect, don’t include typical 1/3/5 triadic voicings in your lowest range. 

Voicing on the left, not the right - better for lower ranges
It's even better to spread out these voicings even more!
My hand just can't reach that far.

3. Use more than one track.  How many tracks you use depends on the context and the ease of workflow.  In any given session, I’ll typically use anywhere from four to eight tracks of just strings.  One is always a “string section” with sustained articulations.  Others may include solo strings, individual string sections with specific articulations, pads, etc.  It all depends on the musical context and the arrangement style (when in doubt of “arrangement style,” listen to temp tracks, or a reference of a film score, orchestral work, gaming score, etc., and pay critical attention to the layering you hear).

4. Layer with more than one sampling program or library.  I regularly even mix in synth pads to blend in harmonies, typically with slightly adjusted voicings, although I will also double many notes in chords to reinforce harmonies.  (It all depends on the context, of course.)  A good tip I’ve learned is to start out with your most “realistic” library that you’ve got, even if it’s not that high quality.  Then “pad” it with harmonic pads, string emulators instead of samplers, etc., to deepen in the sound.  Just make sure your transient peaks hit at the same time (see #9 below).  If your time/budget allows, hire a session player or two on a REAL instrument to double what you’ve sequenced - that’s how many Hollywood sessions work!


In case you're wondering what sample libraries I use in my string arrangements for Helsott, it's nothing fancy.  I do NOT believe that you need the "latest and greatest" to make quality recordings if you know how to use what you DO own.  I have an older version of Native Instruments' Komplete bundle - the Kontakt libraries get a lot of mileage in my studio, although I tweak them.  I use a combination of software instruments for pads, along with a couple of hardware synths, including my old beater Roland X6 (my touring synth) and my even older Novation Nova II (the weirdo 90s era performance synth, not the table-top version).  Again, I tweak the hell out of patches.

5. Velocity editing is essential to create a true legato feel.  If you step edit, or use a pencil tool to enter pitch information, you’ll absolutely need to adjust EVERY velocity level to create anything even remotely realistic.  You have 128 values to use in MIDI language.  Use the entire range.


Note: no two velocity stems are created equal.

6. Pay attention to the release of notes in chords (your note off messages).  Keep these synced (to a large extent) but if everything is always perfect, it’s not necessarily going to sound “real.”  Too perfect = sounds like a computer.

7. Adjust articulations/style if it’s possible.  Many software and hardware samplers include mappings for tremolo, sforzando, fortepiano, pizzicato, etc.  Automate and adjust accordingly.  Keep in mind that in sustained articulations the timing will differ from sforzando articulations, which typically will contain the bloom of the attack right ON the MIDI on message.  If you don’t have this facility, you can accomplish the same type of thing (with the exception of tremolo and pizz) simply by automating the envelope (ADSR values).  Typically ADSR values are mapped to CC information, so they can be edited within your sequence.


If I'm speaking gibberish with the terms pizzicato, sforzando, etc., definitely check out one of those textbooks mentioned above in tip #1!  This stuff is helpful to know.

 Screenshot from a default cello patch in Kontakt.
Click on pic for larger version - note mapping of articulations!
These are below the range of the instrument - input them as a "note on" message.
I embed articulation messages in EVERY MIDI region, believe it or not!

8. ONE OF THE MOST IMPORTANT TIPS EVER!!!  The bloom of the envelope (your transient peaks in audio-speak) should hit ON the beat, not after, otherwise everything will sound late.  Don’t just trust quantization - you’ll need to adjust by ticks/samples depending on the sequencing program you’re using.  I’ve been known to adjust note on messages by as much 30-40 milliseconds with slow, soft, sustained passages.  It’s best to go through phrase by phrase to make these adjustments - what works in some phrases won’t work in others depending on the articulations you’re using.  One tip, however - save a version of your MIDI file BEFORE doing this editing with hard quantization values (everything ON the grid), and export as a type 1 file (the type that preserves discrete track information).  This can be super useful if you switch over to another sampling library, another sequencing program, or if you ever end up working with REAL string players...you can import MIDI files into notation programs like Finale and Sibelius.  On a side note in terms of quantization, I almost NEVER use straight 100% quantizations in order to create a more “human” feel, although I will include a hard quantized version of any session I do in order to export accurate MIDI files to Sibelius (my current notation program of choice, as it syncs with the version of Pro Tools that I’m on!).  Look for “randomization” or “humanize” functions in your sequencing software to accomplish this.

9. When it comes to mixing sampled strings, I will almost always export each track as a separate audio track and then mix in my DAW of choice (Pro Tools).  Do NOT rely on MIDI automation in the mix context.  Yes, 128 values seems like a lot, right?  Nope.  It’s not, not in terms of resolution!

Hope these tips help you take your “string writing” to the next level!  Questions?  Comment below, or ask me on facebook/twitter!

Here I am, having fun on stage with Helsott, on the Paganfest America tour.
photo courtesy of Stellavor Images, ©2013

No comments:

Post a Comment