Social top


Harald Bode’s Frequency Shifters and Vocoders

Originally published in Contemporary Keyboard magazine (February 1980, p. 86) in Tom Rhea’s “Electronic Perspectives” column. Reprinted in eContact! 13.4 with the kind permission of the author.

Harald Bode’s career spans three distinct eras in electronic music instrument design. His Warbo Formant Organ [also discussed by Rhea in this issue of eContact!] came toward the end of a decade of exploration in Germany during the 30s. His Melochord [also discussed by Rhea in this issue of eContact!] found a place in the first electronic music studio in Cologne, Germany, in the early 50s. Now this spry septuagenarian builds sophisticated rack-mount equipment such as the frequency shifter and the vocoder. Bode’s current activities include his own sound company and a relationship with Moog Music, whose frequency shifter and vocoder are constructed under Bode license agreement.

A frequency shifter is a powerful tool in the transformation of acoustically or electronically produced audio signals into new tone colours. Applications range, from simple detuning of quasi-pitched sounds such as drums, bells, chimes and human voice in order to create ambience, or liveness, to the voltage-controlled frequency shift of typical synthesizer waveforms to create new sounds with consistent tone colour over the keyboard range.

In the latter application, a keyboard controller feeds into the control input of a voltage-controlled oscillator as well as the control input of the Bode Frequency Shifter, whose Scale switch is placed in the Exponential mode. The audio signal of the VCO is routed into the signal input of the Frequency Shifter, and many clangourous or bell-like sounds may be produced that stay in tune over the entire keyboard.

Let’s take a look at how these bell sounds are made from our traditional waveform, say a sawtooth pitched at 100 Hz. The sawtooth has harmonics / partials that are exact whole number multiples of the fundamental:

Original wave at 100 Hz:
100 200 300 400 500 600 700 800

Let’s shift the pitch upward 50 Hz. The frequency shifter causes a linear, additive shift; the fundamental is now 150 Hz — but the upper partials have also been shifted 50 Hz, so they are no longer whole number multiples of the fundamental. That is, their ratios are no longer in the natural harmonic series.

Frequency-shifted wave:
150 250 350 450 550 650 750 850

Isn’t there something that transposes the pitch without changing the harmonic structure, and therefore the tone colour? Yep, technically it’s called a pitch shifter. Most people use the term Harmonizer instead, but Harmonizer is a trademarked name for the pitch shifter made by Eventide Clockworks. The pitch shifter and the frequency shifter are sometimes confused, but they do very different things. A pitch shifter preserves the ratios among the partials through a nonlinear (multiplicative) shift:

Pitch-shifted wave:
150 300 450 600 750 900 1050 1200

As you can see, a pitch shifter transposes pitch, but without transforming the tone colour. A frequency shifter changes the pitch for the express purpose of changing the tone colour.

A vocoder is a device that analyzes the tone colour characteristics of a sound as it changes in time and transfers those characteristics onto another sound. The familiar example involves analysis of speech, and the transfer of speech onto the pitched sound of a musical instrument — the “talking instrument”.

Here’s how it works. The program signal — in this case, speech — is sliced into channels by a set of bandpass filters. The amount of energy in each channel is sensed by an envelope follower that produces a control voltage proportional to that amount; we now have a set of control voltages that comprise a coded version of the frequency spectrum of the speech signal.

Another signal, the carrier, is passed through a second set of bandpass filters, where it is sliced into channels as was the program (speech). Typically this signal might be from, say, an electronic organ. Each channel of sliced organ sound is waiting to pass through an individual voltage-controlled amplifier.

Now if we get the control input of each VCA hooked up to the control voltage produced by the corresponding analysis channel, then when the speech signal is rich in highs, so will the organ signal be, and vice versa. If you say “hello” into the microphone, the organ will say “hello”, because speech is made up of lows and highs that change in time!

The vocoder’s association with speech possibly arises because that instrument was first developed at Bell Labs to enhance transmission of telephone conversations. Homer Dudley developed the vocoder in the late 30s because the coded version of speech reduced the bandwidth necessary to carry a conversation; this made room on the wire for more conversations.

Most of the modern designers of the vocoder have followed Dudley’s concepts, even to the point of including a frequency follower to extract and code pitch information in speech. Bode thought that the requirements for a vocoder for “entertaining” purposes might allow departure from those required for speech transmission. He omitted the pitch extractor in his design and made several clever design innovations that earned him a patent.

Central to these innovations was Bode’s realization that the difficult-to-manage highs in human speech (“s” and “f” sounds and other non-pitched sounds) had created havoc in vocoding, causing complexity and attendant expense. His solution is simple, but elegant: he provides a bypass that allows the very highest parts of speech to bypass the vocoder and be mixed directly back into the final output. If you are the phone company in the late 30s this wouldn’t make any sense, for you want to transmit only coded material. But for music, the idea of mixing coded and non-coded material makes beautiful sense! After all, the highs of non-pitched consonant sounds greatly add to the intelligibility of speech; and the coded (normal vocoding) information provided covers the limits of most musical instruments — up to about 5000 Hz.

Oh well, it’s no more than we would expect from a man who designed a valid four-voice (assignment) instrument in the late 30s [the Warbo Formant Organ] and had his instrument in the Cologne Electronic Music Studio!

So far, only the obvious “talking instrument” aspect of the vocoder has been exploited. But there are other possibilities! Play a single violin into a microphone and feed this into the program (speech) input. Route a polyphonic synthesizer into the carrier input. Now play chords. The interaction will produce the polyphony of the synthesizer with the articulation and tone colour of the violin. Or route a rhythm unit into the program input and a polyphonic synthesizer into the carrier. Fascinating “pads”.

The vocoder remains to be fully used musically. Much of its misuse results from ignorance about the simplest concepts of its operation. For instance, providing a rich program signal will avail nothing unless you have a harmonically complex signal in the carrier to be acted upon.

Social bottom