Moog Monday - On Synthesizers: Vocal Sounds, Part III: Practical Simulations

Image placeholder title

(This article originally appeared in the June 1978 issue of Contemporary Keyboard magazine.)

In my April '78 column I listed the array of synthesizer functions that would have to be assembled in order to simulate all speech sounds: a voltage-controlled oscillator, a noise source, four programmable filters, at least one voltage-controlled amplifier and contour generator, and a fairly elaborate sequencer. With this setup, one could theoretically program the production of consonants as well as vowels, and then string them together into syllables and words like so many beads on a string. Alternatively, one can employ a vocoder to impart speech contouring and articulation to harmonically rich steady tones, as I discussed last month.

For live performance use, or for producing occasional vocal timbres, full-blown speech synthesizers are neither necessary nor practical. With a variable synthesizer, a parametric equalizer, a graphic equalizer, and a little ingenuity, you can produce a wide range of vowel-like timbres, as well as a few consonants. What's more, the skills you develop in finding and using vocal tone colors will enable you to more effectively get at other acoustically complex timbres.

Most vowel sounds are characterized by three resonances (see my April column), whereas the typical performance synthesizer has only one resonant filter. A parametric equalizer with two or more sections is the ideal tool for adding frequency response resonances to your synthesizer's audio output. Each section of a parametric equalizer is a resonance whose center frequency, bandwidth, and height (strength) of resonance are independently variable with the panel controls. In addition, a graphic equalizer is useful for tailoring the overall frequency response. Feed the synthesizer output through the parametric, then through the graphic, and finally to your mixer or amp.

Begin by setting both equalizers flat. If your synthesizer has a multimode filter, set it in the lowpass mode. Turn up the Resonance or Emphasis control on your synthesizer's filter, and adjust the filter frequency to about 3kHz. In most synthesizers, setting the filter frequency is easy to do by ear. Disconnect the keyboard control of the filter, play a G 3 1/2 octaves above Middle C, and adjust the filter frequency (sometimes called Cutoff) so that the filter picks out just the oscillator tone's fundamental. The synthesizer filter will be used to produce the F3 resonance. This resonance is the highest vowel resonance. Its frequency is near 3kHz for most vowel sounds.

Two of the parametric equalizer sections will be used to produce the F1 and F2 resonances. For instance, for the "ah" vowel sound, set the first parametric section to 750Hz center frequency, 1/4-octave width, and +15dB height. Set the second section to 1,100Hz center frequency, 1/4-octave width, and +15dB height. The composite frequency response of the synthesizer and the parametric equalizer now contains three resonant peaks that simulate the resonant properties of the vocal tract.

For the initial waveform, you should select a moderately narrow rectangular wave (about 25% duty cycle). If the rectangular waveform width control on your synthesizer is continuously variable, set this control halfway between square wave (hollow sound) and the very narrow rectangular wave (nasal sound).

Attack and decay times are important in vocal sounds. The loudness contour should have a relatively long attack time—about 0.2 seconds. The decay time should be somewhat shorter—about 0.1 seconds. Use no filter contour. Vibrato may be applied to the oscillator. Its speed should be slower than instrumental vibrato—4 to 5Hz at most. Use delayed vibrato if your axe is so equipped. Do not apply modulation to the filter.

With these panel settings, you are ready to "play" the sound "ah." Manual and foot control technique are important. Lift your keyboard hand between phrases or even between notes. Notes should start off soft, build up in intensity, and then taper down. Try playing successive, closely spaced notes by holding a key and varying the pitch with your pitch bender. Vibrato should also be continuously varied, starting with no vibrato at the beginning of a note and building in intensity, as the note gets louder. Remember that a singer produces tones as part of his breathing process. Abrupt articulation and pitch changing are part of the keyboard sound, but they have no place in vocal simulation.

Once you have developed your hand and foot technique, try experimenting with the filter and equalizer settings. The parametric settings given above are a good starting point for an "ah" timbre. For an "ee" quality, set the second parametric resonance frequency at 2.3kHz. Set the first resonance frequency at about 300Hz, and the height about -10dB. This is actually a dip in the frequency response that thins the sound out. For an "oo" sound, try setting the first parametric resonance at 300Hz frequency and +20dB height, and the second resonance at 900Hz frequency and +10dB height. Use the graphic equalizer to get the correct tonal balance. You may not achieve an exact human quality, but you will discover many musically useful vocal qualities.

Here is a technique for producing a "singing soprano" quality that is completely different from that described above. You can use this if your synthesizer has a filter that (a) can be made to oscillate and (b) tracks the keyboard with some accuracy. First set your tone oscillator so it operates in the soprano range. Use a triangular or square wave output. Set your filter feedback so that the filter oscillates. Set the filter frequency so that the filter tone is exactly an octave above the oscillator tone. With the tone oscillator signal going through the filter which itself is oscillating an octave higher, the filter will probably phase-lock onto the oscillator tone, thereby producing a perfect, strong second harmonic. Apply a very little vibrato. Check to see that the oscillator and filter stay locked together throughout the pitch range that you want to play. If they don't, then retune the filter. Now, use your synthesizer volume control to articulate the notes. That is, turn the volume control up at the beginning, and off at the end, of each note. This manual articulation, plus the strong second harmonic generated by the oscillating filter, will produce a warm, vibrant vocal quality.

For more articles by Bob Moog, please visit