In this Logic Sound Design tutorial, we'll be recreating the opening bass sound from Air's "Sexy Boy":
We're referring to the very first sound you hear in the video, which is actually a composite of two sounds, an electric bass and ... something else. Let's get started.
Launch Logic, and start with a new empty project containing one empty Software Instrument track, which will default to the standard EVP88 instrument. With that track selected, select the Library tab on the right side of the screen, choose: 03 Bass > 02 Electric Bass to display the available electric basses.
As I mentioned, the track's bass sound is two layers, one a straightforward, fairly dark electric bass (played with fingers rather than a pick, it sounds like), and the other a synth. Try loading some different electric basses (by clicking on their names in the rightmost column) to see what they sound like - to my ears, the "Subby Bass" works well for this track, but "Liverpool Bass" or "Motown Bass" could also work. I'm going with the Subby.
Here are the first 8 bars of the track, at a tempo of 112 bpm, in both Score and Piano Roll notations:
(These 8 bars play three times at the beginning of the sound; on the fourth repetition, the second half of the part moves up to an F chord. For the purposes of this tutorial, you can just record and loop the first 8 bars while you work on the sound.)
Well, that was easy. The second sound is a little trickier, however. First, assuming you've recorded the bass notes in the first track, create another new Software Instrument track, leaving it at the default EVP88 for now. Double-click the first track's region to open the Piano Roll editor, then click in the Piano Roll editor pane to make it active. Type Command-A (Select All) to select all the note events in the editor window, then type Command-C to copy them all to the clipboard. Select your second track by clicking on its name in the Arrange window, position the Playhead at the beginning of the first track, and paste (Command-V) the copied MIDI data onto the second track.
Hit Play, and you should hear the original bass part doubled with the electric piano - not a bad sound in itself.
The second sound that doubles the electric bass has a guttural vocal quality to it that derives from what are called formants. Formants are the particular multiple resonances or filter peaks, created by our mouth, throat, and head cavities, that define the sound of vowels in human speech and singing. "A, E, I, O, U", and other open vocal sounds all have a unique set of fixed resonances'"usually three'"that let us tell them apart. Those actual resonant frequencies have been well-known to acoustics researchers for years, and they can be applied to synthetic tones to simulate vowel sounds.
Here's one list I found that shows the formant frequencies (expressed in Hertz) for some common vowel sounds:
In our case, the track's basic synth sound is a pretty straightforward sawtoothy waveform ; it's the formant filtering that gives it that vocal quality. So, we need to find a Logic synth or filter plugin that will let us set multiple filters to the necessary vocal formant frequencies, and then vary them to create the different vocal sounds in the track. (There's a free VST plugin called Delay Lama that is designed to do just that. It might well have been used in the original track, in fact. Of course, Logic won't load VST plugins....)
However, after much experimentation, I was unable to find any Logic synth or filter plugins that would let me set them with the precision necessary to create the formants. As you can see from the list, the specified frequencies are fairly close together, especially in the lower range, and even a slight variation in the formant frequency will change the sound significantly. The ES 2 synth, while it does have a good bandpass filter, has only one (so we'd need three instances of the synth to get the necessary three frequency bands), and even then, its cut-off frequency is set in a range from 0 to 1, rather than in absolute frequencies, making it impossible to set the exact values needed. The Parametric EQ plugin, which does let you set its cutoff in terms of frequency, isn't precise enough to let me choose a value of 1190 Hz, for example. This applies to all the other native Logic filter plugins as well. Even the EVOC 20 PS Vocoder synth, and its associate EVOC 20 Filterbank plugin, which do have formant filters, don't let you set those accurately enough to get specific formants.
So, to get close to recreating this sound, we'll need to use a 3rd-party plugin. Fortunately, there's a free one available from the good folks at Music Unfolding called OttoPhormant which, while not perfect, will let us get quite close to the Sexy Boy sound. You can download the Audio Unit version here.
Follow the instructions in the manual to install it (you just drag it into your Components folder in your system or user Library), relaunch Logic to load the plugin, and let's move on.
First, of course, we need to have some kind of sound to apply the plugin to. As I mentioned, a simple sawtooth is all we really need as a source for formant filtering, so instead of the default EVP88 instrument on our second track, load the ES 1 synth.
Here are the ES 1 settings I'm using to produce a straight, unfiltered sawtooth wave:
Note: You can actually use any synth instrument you want that will produce a straight sawtooth, such as ES2, but ES1 is quick 'n' easy to set up, so I'm using that.
Our next step is to apply OttoPhormant to the sawtooth track and get the basic plugin settings in place. First, select the sawtooth track and click in the first Insert slot in the Channel Strip to open the plugin menu, then choose Audio Units > MusicUnfolding > OttoPhormant > Stereo to load the plugin. I encourage you to read OttoPhormant's brief manual to get an idea of how it works, but here's a screenshot of my basic settings:
To set the values for the various properties in the plugin, you either click on the property name, so that green bars appear on either side of the property block to indicate selection, then drag the big central knob to set the value, or in some cases click directly on the value to open a popup menu. You can also double-click on many properties to enter values numerically.
Most of the values you see here actually don't matter: OttoPhormant is intended to automatically morph between the different formant sounds using internal envelopes or the level of the input sound. In our case however, we're going to automate the formant changes in Logic to precisely sync them with the track. So, the only values that really matter here are that Envelope Type needs to be set to None, the Filters type should be set to Three Stage BP, Formant Filter Drive should be 0.6, and Formant Filter Resonance should be 0.75. (As always, you should experiment with these settings to hear what they sound like.) You also want to be sure your Dry Level is set to -80 dB, so we don't hear any unfiltered sound. If you play your track now it should already be much closer to the Sexy Boy sound.
Our next step is to automate the changes in the formant filters so that the changes sync with the notes in the track. If you listen to the original again, you'll hear that the low D and C are very bright and buzzy, but still with a certain vocal quality, while the upper D has something like an "A" sound. The upper C that your hear slurring into the upper D, by contrast, sounds kind of like an "OO". Those sounds are available as formant settings in OttoPhormant, so all we need to do is add some track automation to switch between those formants at the appropriate times.
First, of course, we need to show Track Automation to work with it, so choose View > Track Automation (from the local Arrange menu), or just type A. You'll see the Track Automation area open up in the note region for each track, with Volume chosen as the default automation type, an automation date line with the current Volume value in yellow, and below that, the MIDI notes in the track shown as gray bars below the automation info:
Next, select your sawtooth track, click on the popup that currently says Volume, and choose OttoPhormant > Start Formant as the parameter you want to automate. Because we aren't using the envelopes in OttoPhormant itself to change the formants, the Start Formant is the only one that you hear and the one we want to change. Now we're ready to start automating.
Next, make sure you have the Pointer Tool selected from the Tools menu in the upper right corner of the Arrange window, then click on the very first data point at the beginning of the automation line and drag the point down until it reads "I" - which is the formant we want to start with.
I chose "I" for the formant because I thought it most closely matched the buzzy low note in the original track, but you're welcome to try others if they sound better to you.
Next, line your Pointer tool up with the beginning of the second MIDI note in the track, the high D, click to create a second automation node, and drag it up to set a new formant value - I've chosen the "E" formant, because I think it sounds most like the "A" sound from the original track:
Next, line your Pointer tool up with the beginning of the third MIDI note in the track, the low D again, click to create a third automation node, and drag it back down to "I". Position your cursor over the fifth MIDI note, the high C (the third and fourth notes are both low Ds), and click and drag to set it to "U". Click over the sixth note and set the formant to "E", and set the seventh note's value to "I" again. Your automation track should look something like the image below.
Note: Unfortunately, OttoPhormant seems a little flakey about how it handles automation data - you'll probably have to tweak the positioning of the automation nodes to get them to line up correctly with the note changes, and I've noticed some inconsistency in the automation playback. But hey, it's a free plug-in!
At this point, you can continue manually placing automation nodes to complete the automation for the track. One shortcut you can use to make that all a little quicker is to duplicate the existing automation data further along into the track. To do that, choose the Automation Select tool from the tool menu in the Arrange window.
Drag it over the automation data you want to copy. Then Option-drag on the last node in the data group to move and copy the data group later into the track:
Once you've copied the automation data further down into the track, you can drag the entire selected group to exactly line it up with the correct MIDI notes.
You can hear that there are still a couple of issues with this recreation: one is that there should be a smooth transition between the "OO" and "A" formants on the upper notes in the track, but unfortunately OttoPhormant only allows abrupt transitions between the formants when using automation. Internally, when using its own envelopes, you can get a smooth morph, but not externally. Still, that part of the recreation is pretty close, if not exact.
The other issue is that the first note, the low D, is a lot buzzier than what we have here. So, for a final touch, I'm adding one more track to double just the low D, using distorted sawtooths (sawteeth?) and pulses for the sound. I'll leave that last little exercise to you:
And there we have it! As always, I encourage to experiment, and as always, have fun.
Interested in funky, filtering and formanting? You might like David Earl's Vocoding With Logic!