Project 4-1: Listen to the Waveforms. To do this project, you’re going to need to create a default patch for your synth. Start by using the “initialize voice” command if the instrument has one, but be aware that the init voice may already have some types of processing (such as a tasteful amount of reverb) that you’ll need to remove. The goal is to create a patch that has the following characteristics:
• Only one oscillator should be heard. You may be able to do this by turning the others off, or by turning their output levels down to zero in the mixer.
• The filter should be wide open. You may be able to shut it off entirely. If it’s a lowpass filter, crank the cutoff frequency up to the max, and turn the resonance down to zero.
• The amplitude envelope should be set to instant attack and full sustain level.
• There should be little or no velocity-to-amplitude modulation: All the notes you play should sound equally loud.
• The effects should be switched off or set to 100% dry.
• The oscillator should have no modulation inputs. Either switch off the modulation, or turn the levels down to zero. Some synths hide their modulation routings in odd places, so you may have to hunt for them.
After creating this patch (and saving it as “MyInitPatch”), you’re ready to proceed. Call up the relevant edit page of the active oscillator in the display and go through the synth’s waveform selections one by one. Listen to the sound quality of each wave. Play them near the top and bottom of the keyboard, not just in the midrange. If your synth has velocity cross-switched multisamples, play at low, medium, and high velocities to hear the different samples.
This project has a dual purpose. First, you’ll be getting acquainted with the full range of raw sounds available in your synth. Second, you may find that some of the waveforms don’t sound very good in their raw state. Filtering, enveloping, keyboard scaling, and other techniques are needed to produce musically pleasing sounds. Once you know the raw materials, you’ll be able to apply those techniques more intelligently.
Project 4-2: Combine Waveforms. Beginning with the default patch you created in Project 4-1, activate a second oscillator. Make sure the second oscillator’s detune or fine tune control is set to 0 (no detuning).
Try the following experiments:
• Listen to two different waveforms mixed at the same loudness level. If your instrument has 100 waveforms, there will be 10,000 possible combinations with two oscillators, so you may not have time to listen to them all. Let your intuition be your guide.
• Try some of the same combinations of waves with the second wave tuned an octave or two higher (or lower) than the first.
• Try some combinations with the second wave mixed at a very low level, so that it adds a barely perceptible color, or filtered so that only its lowest or highest overtones contribute to the composite sound.
• If your instrument allows each oscillator to have its own amplitude envelope, give the second oscillator a very short plucked envelope, so that it adds only an attack transient to the tone. Try tuning this attack transient up an octave or two, and try reducing its loudness.
• Program the second oscillator so that its loudness responds more to velocity than the first oscillator’s. The blend of the two tones should change, depending on how hard you strike the key.
• When you’ve found a combination that seems interesting, refine other aspects of the patch to taste.
Project 4-3: Discover the Waveforms in the Factory Patches. Choose a factory patch that you like. Go into edit mode, and figure out how many oscillators it uses. Listen to each of the oscillators in isolation.
(You may be able to mute and unmute oscillators, or you may need to turn their outputs down to zero to mute them.) Notice the contribution that each oscillator makes to the composite tone.
You may find that some oscillators seem to make no sound. This may be because they’re active only at high (or low) velocities, or because they’re only active in a restricted range of the keyboard.
Project 4-4: Swap Waveforms. Choose a factory preset in your synth. (Almost any preset will work for this experiment.) As in Project 4-3, go into edit mode and determine what contribution each oscillator is making to the tone. Then, without making any other edits, try changing the waveform of one oscillator to see how the character of the tone is altered.
Sometimes, changing the waveform is all you need to do in order to create a radically new and good-sounding patch. If you find something good, save it to one of the empty locations you created in Project 3-1. Other times, you may find a combination that seems promising, but that needs work. Experiment with the tuning of the oscillators, the envelopes, and the filter to help the new waveform blend in better with the other waves.
Chapter 5
Filters
Once you have an oscillator generating an audio signal, as described in Chapter Four, there are two things you can do with the signal to make it more interesting or more musically expressive: You can add new harmonic content, or you can take some of the existing harmonic content away. Adding new harmonic content is most often performed by effects processors, which are discussed in Chapter Nine, or by means of FM synthesis, which is discussed briefly in Chapter Four. In this chapter we’ll cover the process of stripping away portions of the harmonic content from a signal. Modules that do this, or (to be a little more rigorous about it) that change the relative balance of the sound’s partials without introducing any new partials, are called filters.
If you’re not familiar with the nature of the harmonic spectrum, this would be a good time to review the material on that subject in Chapter Two. Filtering is a frequency-based process — that is, filters differentiate among the various frequencies in an incoming signal — so we’ll be talking about the frequencies in the harmonic spectrum, and how filters treat them. In the discussion that follows, the words
“harmonics” and “overtones” will sometimes be used to refer to the various frequency components of a signal, whether or not they’re harmonically related to a fundamental. This usage is a little sloppy. The word “partials” is more correct, but is less commonly used.
Broadly speaking, a filter will reduce the amount of energy in certain parts of the spectrum, while perhaps increasing the amount of energy in other parts of the spectrum and leaving still other parts of the spectrum entirely untouched. With one exception, which we’ll get to below in the discussion of resonance, filters don’t generate any sound on their own. That is, if the signal being processed by the filter contains no harmonics within a given frequency band, the filter won’t have any effect at all within that band. It won’t create partials that don’t exist in the raw sound. But if there’s a harmonic component, which thanks to Fourier analysis we can visualize as a sine wave, within a band that the filter is operating on, the filter may decrease the amplitude of the sine wave, increase it, or simply pass it through without doing anything to it.
Figure 5-1. The response curve of a typical lowpass filter.
Actually, “without doing anything to it” is an oversimplification. In addition to changing the amplitude
of frequency components, filters can also change their phase. This fact isn’t usually of much significance to musicians who use synthesizers, because, as noted in Chapter Two, the human ear does a poor job when it comes to distinguishing the phase relationships of harmonics within a tone. If a synthesizer filter’s phase response made an audible difference, you’d probably see lots of instruments with filters that included some form of phase control as a feature. For some purposes in audio engineering, such as the design of filters that are used in conjunction with analog-to-digital converters, phase coherence (that is, keeping the overtones’ phase relationships constant) is a real issue. These filters are not discussed in this book.
The operation of filters is conventionally diagrammed in the manner shown in Figure 5-1. If you don’t understand this diagram, you’re going to get lost pretty quickly, so let’s dissect it in some detail. The frequency spectrum is plotted on the graph on the X axis, with low frequencies at the left and high frequencies at the right. The amount of gain (boost) or attenuation (cut) introduced by the filter is plotted on the Y axis. The response curve of the filter is drawn on the graph.
The values on the X axis are plotted in an exponential rather than a linear manner. That is, each octave takes up an equal amount of horizontal space on the X axis. This is not the only way to do a frequency plot, but in the case of filters, it makes sense, as we’ll see.
Figure 5-2. The response curve of a typical highpass filter.
In the simplest case, the response curve would be a horizontal line stretching from left to right at the zero point on the Y axis. In this case, frequency components of the incoming signal would be processed with no gain and no cut, no matter what their frequency. (Believe it or not, this type of filter, which is called an allpass filter, is useful in some situations, such as building reverbs and phase shifters. This is because an allpass filter changes the phase of various harmonics.) At frequencies where the response curve dips below the zero point, the incoming signal will be attenuated, and at frequencies where the response curve rises above the zero line, the signal will be boosted.
How much boost or cut is introduced at any given frequency is shown by how far the response curve is above or below the zero line. If the response curve dips down to a low level, the corresponding frequency components of the signal being processed will be attenuated by quite a lot, rendering them nearly or entirely inaudible. If, on the other hand, the response curve is only a little below the zero line at a given spot, the frequency components of the signal in that part of the spectrum will be attenuated only slightly.
Explaining exactly how a filter can perform this type of magic on a signal is beyond the scope of this book — which is fortunate, because I’m not an engineer, so I have no idea how it works. All I know is, it works. Filters can cut some frequency bands while at the same time boosting others. A filter is a frequency-dependent amplifier.
Figure 5-3. The response curve of a bandpass filter.
Figure 5-4. The response curve of a band-reject (notch) filter.