Make your MIDI sound real! The following is Part 2 of the transcript from Midi Film Scoring’s video tutorial originally titled “Getting the Most out of Your MIDI Sequencer,” authored by film and television composer Dan Brown. Learn how to make realistic MIDI mockups by applying MIDI orchestration concepts, including panning, keyswitches, continuous controller data, tempo, and more. If you haven’t already, be sure to check out our previous tutorial first: Make MIDI Sound Real: Creating Orchestral Mockups, Part 1.
Transcript: In the first video of this series, we covered some basic MIDI orchestration techniques to improve the sound of your MIDI mockups. Now we’re going to delve a little deeper, still working with the same music and sequence we’ve started with.
Now we need to ditch the generic string patch we were using and replace it with five individual MIDI channels and patches for each string section. If you record all your string parts at once on one polyphonic channel, you can accidentally cause your string players to multiply. If we play one note on this patch, we hear 14 recorded strings. Add another, and we double it to 28.
A typical string section has about 50 players, but if you record all your strings on one generic patch, you are going to switch from only 14 players on one note to suddenly hearing 70 on all five string parts. This fluctuation of numbers doesn’t sound quite right to our ears for obvious reasons. Recording the string sections one at a time has the even bigger benefit of giving us what I like to call “string slop.”
A great string section sounds full and lush because it contains so many different players starting and stopping at ever-so-slightly different times, playing notes just barely out of tune with one another. All these tiny discrepancies are what make us love the sound of a large string section, and playing your string parts in one at a time helps give your sequence a little of those timing and tuning differences. Here, we load in a violin legato patch.
Now I’m recording the Violin I part by itself, making sure to use the mod wheel and expression pedal to lend me some human feeling.
[Ensemble playing]
In addition to playing every section in one at a time, we’re going to create a solo MIDI channel for every part and load in a solo patch for every instrument. These solo patches are recorded separately in the studio from the ensemble patches so, when played alongside our existing string parts, they will add another tiny bit of pitch and timing variation. Just be sure to turn the solo channels down about -10dB at the audio channel level, or set controller 7 to around 60, to reflect how loud that solo player would actually sound compared to the section patch if we were listening to this orchestra on a stage.
[Violin playing]
We also don’t use solo string patches to pad out the double bass line. Scary creatures lurk inside the solo double bass, and we don’t want them creeping into our recording.
Our reverb plugin gives us an aural image of the hall we’re hearing the music in, but all the music is coming from directly in front of us. A more realistic sound would have instruments spread out from left-to-right before us. We can accomplish that by panning the instruments in our instrument plugin as if they were spread out on a concert stage. This helps your part stick out more clearly without having to turn up the volume on any tracks.
Consult a diagram of an orchestra stage setup to see where each instrument should be panned to if you don’t already know. Solo instruments and bass-frequency instruments should be kept panned in the center, regardless of where they are on the stage.
Now we’re going to move to our solo melody line on the English horn. Looking at our notation, we can see that our melody uses two different kinds of articulations, or ways of playing. In music, any notes that are played kind of “normally” should be played using our legato patch. The legato patch not only records the notes, but the sound the instrument makes as it changes between one note and another — an important part of recording samples properly that our ear needs to hear. Legato patches aren’t held out for very long, however …
[English horn playing]
… So on our long notes we will need to use the sustain patch, where the library creators have taken the note recorded and looped the recording to play much longer.
[English horn playing]
I’ve also called for some notes to be played very short, or using staccato articulation.
[English horn playing]
So now we’re going to use three articulation patches for this MIDI channel. We’ll default to the legato patch, play long notes on the sustain patch, and the short, quick notes on the staccato patch. We need to be able to switch between these patches quickly while recording. We can do this quite easily in Vienna because of the square patch matrix Vienna allows us to use.
Here you can see that we can drag all the patches we want into the square matrix. You can tell Vienna to use multiple things to switch between these squares on either the Y axis or the X axis. I’ve put all three patches on top of each other, with staccato on the top square, legato in the middle, and sustain on the bottom square. To switch between the three, I will set my pitch bend wheel to control what patch I have selected. Pitch bend is a great tool to use for switching patches because it’s on a spring, so it will jump back to your default middle patch — legato, in this case — as soon as you let it go.
Back in the basic window of Vienna, we can see there’s multiple ways to switch between patches. Not many sample libraries have this kind of box matrix setup or allow you to change patches using the pitch bend but — more commonly — map each patch to a keyswitch on your keyboard that you can press to change patches. Visually, it would be represented this way, with each patch in a row and the leftmost patch assigned to a starter key, like A0, then the next patch assigned to A#0, then B0, then C1, et cetera.
Keyswitches can be harder to navigate in real time while you’re recording, so people often input them on the track before recording so your patch changes automatically while you overdub the notes on top of it. This means that your keyswitch notes will show up on the same track as your musical notes. This can visually confuse people editing music in the piano roll, as well as really screw up your sheet music if you’re the kind of person who prints your music directly from your sequencer.
To get around this, some people will create a separate MIDI channel directly under the first one, assign it to the same MIDI output so it affects the same patch in your sample library, and record the keyswitches onto that channel. This keeps your keyswitch notes visually out of the way of your musical notes.
Personally, I prefer to keep everything in the same track. It is not optimal to create a separate MIDI channel for every articulation you use, however. This creates unnecessary clutter in your sequence and drain on your computing power, particularly if you’re creating a full orchestral sequence.
Also, always be sure you record some pitch wheel or keyswitch data in your bar zero to make sure your plugin plays back the correct patch on the first beat. In the first video, we set the modulation wheel and expression pedal to control our volume and velocity crossfader for our strings.
Here we have a problem, though. We want to have our mod wheel control the dynamic level selection for held-out notes, but on short notes, like our staccato patch, it’s more natural-sounding to allow the velocity to be chosen by key velocity, so we are going to map the velocity crossfade box to the pitch bend wheel so that when we pull down on the pitch bend to switch the patch up to staccato, the box unchecks itself, changing us from the modulation wheel control to key velocity control.
[English horn playing]
Side note: In the newest update of Vienna Ensemble, you need to remember to map the pitch slider in the Advanced Vienna window to a different controller on your MIDI controller, or it will also slide the pitch of your note as you use the pitch bend wheel. If you accidentally move the pitch slider while reassigning it to another controller, you can right-click it twice and it will reset the pitch slider to its default position.
Now we can record our English horn using our three articulation patches.
[Ensemble playing]
It still doesn’t sound right to my ears, though. If we look at the piano scroll, you can see the English horn notes are overlapping with the string notes. This means that the string part is partially covering up our melody. A good composer would anticipate this as a problem when he wrote the music and set the melody up an octave to ride above the string pad.
[Ensemble playing]
If we try this with the English horn, it begins to sound kind of squeaky — it’s not as full-bodied. Part of sequencing well is using instruments in the ranges they are most strongly suited to, and the English horn is pushing its luck at the top end there. The oboe has a very similar sound to the English horn, but it sits in a slightly higher register, so I’m going to replace our current patches with oboe patches.
[Ensemble playing]
This sounds much more idiomatic to our instrumentation, and our melody is no longer obscured by the string lines. When I switch over to the French horn part, I realize that playing the solo French horn doesn’t sound as bold as I would like …
[French horn playing]
… So I’m going to change the patch to an A4 patch, or a patch that has four horns recorded playing the same note at once.
[French horns playing]
If you know your sample library front to back, you will know which instruments and patches serve which situations best.
Now let’s say you made a mistake while recording. It’s often faster to go into your MIDI channel and edit your mistake than to re-record the entire part. If we pull open our MIDI channel, we can see our notes displayed in the piano roll on the top, and all the MIDI data displayed at the bottom. Here, I can select which data to edit through this drop-down menu and then draw or erase it using my DAW’s tools to correct any errors. Something I frequently have to do is smooth the string fades or change the MIDI data to make a dynamic swell a little more accurate.
Many sample libraries don’t have a fader for the controller 7 volume setting, so you can set it here in your MIDI track by selecting “volume” from the controller drop-down menu and drawing one point of data into your MIDI channel.
If you are dealing with a velocity-controlled instrument, like the staccato notes of our English horn, we can drag the velocity to where we need it and even move our pitch bend or keyswitch data in case we didn’t get the patch change just right. MIDI data is handled differently in every sequencing program, so learn how to mess with all of these things regularly in your favorite DAW.
Something heavy piano roll users like to do is quantize their notes, which makes any selected notes snap to your defined grid point. Orchestras are very precise in their timings on medium-to-fast music, but the slower and more textural your music, the less you should quantize it. If you hear audible mistakes or it sounds sloppy, correct it, but, as a general rule, don’t quantize your entire piece of music, particularly the strings. All the other sections can be quantized a little more heavily, but rarely quantize above 90% strength, unless the music is fast-paced.
The last thing we’re going to fix is our tempo. Please note that you can slow your sequence tempo down while recording to make it easier for you to play the notes in if you need to. After recording, put the speed back to where you want it, and then start fine tuning.
Right now, everything sounds pretty good, but our music is being played at the exact same speed all the way through. Even if your music is set to the same tempo all the way through your piece, human error will vary the tempo by several beats per minute at any given time. This means it’s a good idea to randomize the tempo by about 3% for your entire piece using your software’s tempo change function.
Looking at the tempo track, we can see it now varies in tiny amounts. I don’t want approximately the same speed throughout this piece, however. This isn’t an action cue, and it could benefit from some small tempo changes at the ends of the phrases to really give it some extra emotion.
I’m going to select my pencil tool and draw in some ritardandos, or tempo decreases, at these crucial points in the music, and I’ll draw in a slightly slower tempo over the horn section to create a more majestic feeling.
[Ensemble playing]
It is important to do this step last, or you will have a much more difficult time recording your music in accurately while the tempo changes. These changes in speed are the icing on your musical cake, allowing your music to sound like a live symphony orchestra.
Finally, we’ll play back first our incorrectly-sequenced version.
[Ensemble playing]
And then, our final product.
[Ensemble playing]
There’s a world of difference between the two, and all of it done with a single sample library and a MIDI controller.
If you paid attention today, you learned about:
- “String slop”
- Panning
- Changing patches and articulations
- Keyswitches
- Knowing your sample library front to back
- Editing controller and MIDI data
- The dangers of quantizing orchestral music, and
- How to edit tempo
Remember, it’s not about having the most expensive sample libraries out there, but how you use them, that really matters.
[Ensemble playing]
Transcript by Calgary Transcription Company. Featured image by Steve Snodgrass (CC BY 2.0).
mike says
I have east-west orchestral samples. Their section of violins to me don’t sound like the real thing even with reverb. It does not have the scratchy sound that the bow gives strings. Is there a way to make a section sound like the real thing?
I have noticed that recording engineers don’t pan orchestral instruments the way they are arranged on the stage. All instruments sound like they are coming from left and right. For instance , the violins are on the left side and the right. Strings basses are panned right and left and same with all instruments.