Recording - June 2017 - 72

FadeOut
THE COMPUTER THAT CRIED "DISCO SUCKS"
Guest editorial by Scott Dorsey

Two important things happened in the early 1970s: small
and limited but usable microcomputers suddenly became available
to the general population, and dance music began a rush out
of the clubs and onto mainstream radio. Microelectronics-
in the form of microprocessors and electronic synthesizers
(combined with primitive digital sequencing), both advancing
rapidly-made for these two revolutions, which we will call the
Microcomputer Revolution and the Disco Revolution.
Like all revolutions, both of these movements caused counterrevolutions from those who were already fully involved with existing
music and technolology. Thus by the late 1970s we began to see
reactionary T-shirts like "If It's Not 36 Bits You're Not Playing With
a Full DEC" and the simple but elegant statement "Disco Sucks".
There are historians today who claim that the Disco movement
and the corresponding Disco Sucks movement were motivated
by many different factors-social, economic, artistic, even racial.
This might be the case, but from my perspective at the time, it
looked just like an argument between two different groups of
people, both of whom had lousy taste in music.
This said, let us go back to the distant era of 1979, when
a small basement company on the West Side of New York
called Eventide Clockworks came out with its THS224 third-octave
spectrum analyzer. A number of companies at that time, like
GenRad and Brüel & Kjær, had made computer-controlled
spectrum analyzers before, but the Eventide folks had done
something different. The THS224 was an inexpensive card
that just dropped into a Commodore PET microcomputer with
a programming interface. Using the BASIC programming
language, the PET could set gains and take data from the card.
Now, this was long before there was enough computing power
available in a microcomputer to just digitize the audio and use the
CPU to do a Fast Fourier Transform analysis, the way we would
analyze audio today using an audio interface and a program like
SpectraFOO or Matlab. The card was actually a hybrid device
with a row of 1/3-octave filters whose outputs could be digitized;
all the filtering was being done in the analog world. The massive
advantage of this approach is that it allowed almost all of the CPU
power to be used for analysis of the third-octave data.
This was state of the art! A number of the companies that
started out building hybrid digital-controlled analog devices in
that era managed to make it big and stick around for many
years. Novation, now known as a manufacturer of controllers,
got its start at about this time with the Apple-Cat, a modem
that was a small basic synthesizer under CPU control. As a
general-purpose analog signal device under computer control, it
was able to work with any arbitrary data format, and could
generate and detect a wide variety of non-modem tones as
72

RECORDING June 2017

well as modem data. It found great interest among amateur
radio operators and phone phreaks for that reason.
That being the case, with a lot of early hybrid signal
processing gear starting to come out in the early microcomputer
age, Eventide was pressed to find some way to distinguish
and publicize its product. One of the ways the firm did this
was by sponsoring an applications contest, in which contestants
could win an analyzer card by submitting a program that
used it in a clever way.
My recollection is that Eventide showed the winner of the
applications contest at the AES show in 1979 or 1980. This
program looked at the lower few third-octave bands, looking
for repetitive signals. In other words, it was looking for narrow
frequency peaks in the envelope of the signal within each one
of the bands (what today we would call modulation domain
analysis). Basically, it was listening for a beat. If it found one
that had the audio characteristics of a four-on-the-floor kick
drum at around 120 BPM, it would immediately sound an
alarm, displaying the message "DISCO SUCKS".
As far as I can tell, this integrated system of hardware and
software was the first device to effectively perform accurate
blind music categorization. By "blind" we mean that no human
intervention is taking place; the system is listening to audio
and then categorizing it.
Today, in this age of enormous music libraries and iTunes,
automated music categorization has become a field unto itself.
Even if the state-of-the-art algorithms are far from 100% accurate,
things have progressed considerably in recent years. But it all
started out with a Commodore PET sitting on a trade show floor
with the message "DISCO SUCKS" flashing on the console.
Full disclosure: I showed this article to Richard Factor, one
of the original founders of Eventide. He said that they held a
contest for the best program to distinguish disco from other
forms of music, but that they got absolutely zero entries. He
also pointed out that at the time, Eventide's facility was directly
across the street from Studio 54, a club that was to disco what
Memphis was to Elvis or Chicago was to organized crime,
and they certainly wouldn't have been flashing a display that
said DISCO SUCKS. It's interesting, then, that both my fellow
Recording author Mike Rivers and I have this impression in
our heads of seeing a PET machine flashing DISCO SUCKS
on the trade show floor....
Scott Dorsey (dorsey@recordingmag.com) is an audio
engineer, electrical engineer, and historian of the odd and
obsolescent in audio and computing and probably a few
other fields, living and working in Williamsburg, VA.

Table of Contents for the Digital Edition of Recording - June 2017