Demise of Fairlight Instruments.

[munnari!augean.oz.au!ross%chook@uunet.UU.NET (Ross Williams)]

>Apparently Fairlight Instruments, makers of the Fairlight CMI (a truly
>phenomenal music workstation, most recent incarnation the Series III)
>have gone out of business as of about 2 weeks ago.

>The phone rings unanswered at the New York office, the LA number has
>changed and there's only an answering machine identifying the number
>as belonging to Fairlight and asking people who need service or have
>tech support questions to leave a message, and the Nashville office
>seems to have disappeared altogether.
>
>Anyone heard anything else?

The following message was posted in a general newsgroup in Australia and
may be of help.

Ross Williams.

+----------------------------------------------------------------------------+
| Ross Williams,                ACSNET, CSNET, BITNET : ross@comsci.ua.oz    |
| Dept. of Computer Science,    ARPA   : ross@comsci.ua.oz.au                |
| University of Adelaide,       UUCP   : ...!uunet!munnari!comsci.ua.oz!ross |
| G.P.O. Box 498,               Phone  : +61 8 228-5585                      |
| Adelaide 5001,                Fax    : +61 8 223-1206                      |
| South Australia.              Telex  : UNIVAD AA 89141                     |
+----------------------------------------------------------------------------+

-------------------------------------------------------------------------

Path: augean!hugin!mimir!munnari!basser!softway!andrewb
From: andrewb@softway.oz (Andrew Bettison)
Newsgroups: aus.general
Subject: The Decline and Fall of Fairlight Instruments (LONG)
Keywords: music, technology
Date: 17 Mar 89 05:22:13 GMT
Distribution: aus
Organization: Softway Pty Ltd, Sydney, Australia
Lines: 208


           THE DECLINE AND FALL OF FAIRLIGHT INSTRUMENTS


Disclaimer:  this article is based entirely on my experience as an R&D
programmer with Fairlight Instruments for the last 3 years.  I make no
claim to more knowledge of the details of events described herein than
any other of the (ex-) Fairlight R&D staff.  I just know what we were
all told.

Fairlight Instruments, as you may know, were a wholly Australian
company.  Their products were the Fairlight CMI Series I, II and III
(Computer Musical Instrument), the Fairlight CVI (Computer Video
Instrument) and the Fairlight Voicetracker.  January 1989 saw the end
of their ten-or-so year history of technical and musical innovation
which had made them one of the most truly amazing things to happen to
music in this decade (flames welcome).

It all started back in the late seventies when three enterprising
fellows, Kim Ryrie, Peter Vogel and Tony Furse, collaborated on making
an entirely new kind of musical synthesiser.  Lore has it that initial
technical development took place in Kim Ryrie's grandmother's basement
which had a view of Sydney Harbour.  Whilst searching for a name for
their new product and company, the Fairlight hydrofoil hove into view,
and thus was born Fairlight Instruments.  (The Fairlight hydrofoil has
since been decommissioned by the UTA, and two of its life savers were
presented to Kim and Peter at the end of 1987.)

The Fairlight CMI (Series I) was built around a pair of M6800s and
eight channel cards, each of which held 64K of Waveform RAM and could
produce a single monophonic voice by sending the contents of its WRAM
through an 8-bit Digital-to-Analog Converter and applying
digitally-controlled analog post-filtering and gain.  Polyphony was
achieved by loading a waveform into several channel cards and mixing
their analog outputs.  Waveforms could be created using a technique
called "sampling", which involved converting an analog audio signal
into digital waveform using an Analog-to-Digital converter.  Sample and
playback frequencies of up to 35 kHz were possible, providing an
effective audio bandwidth of up to 17 kHz.  The 8-bit resolution of the
samples represented an equivalent dynamic range of 48 dB.  Essentially
here was an instrument which could reproduce almost ANY natural sound
with convincing clarity.

Software was written to munge existing waveforms or create new ones
using windowed FFT synthesis.  The channel cards were capable of
"looping" portions of waveforms to create sustained voices and provide
a crude but effective time dilation facility (altering the pitch of a
sound without altering its duration or amplitude envelope).  An eight-
part rhythm sequencer, based on the Linn drum machine, was created and
enhanced to incorporate melody, dynamics and note durations.  The
unit's dual 8" floppy disk drives allowed waveforms and songs (in the
form of rhythm sequence files) to be stored and retrieved.  The MCL
(Music Composition Language) allowed composers to compose music
algorithmically and with mathematical precision.  Fairlight Instruments
created and distributed a "sound library", the voices of which can be
clearly heard in most major pop records produced in the early eighties,
and are now considered `classics', available on most sampling synths.

Peter Vogel went on a world tour, lugging the first Series I from
studio to studio in search of interested musicians.  Stevie Wonder was
interested, and since then the Fairlight CMI has risen to legendary
status as one of THE musical instruments of the modern age.  (Stevie
has since taken to using New England Digital's Synclavier - the
(extremely expensive) product of Fairlight's major competitor.)

The CMI Series II was essentially identical to the Series I, but used
M6809 CPUs and had better architecture.  The whole thing fitted inside
a sturdy frame about 4' by 2' by 1.5' and was remarkably resilient
(roadie-proof).  They went for about $35k each.  More than a thousand
Series IIs have been sold worldwide (I think).  Production of the
Series II stopped in 1985 or thereabouts.

The CVI provided a dazzling selection of low-resolution (255 x 255)
realtime video effects controlled from a simple, intuitive array of
sliding controllers.  It sold for around $7k (Aus) and was by far the
cheapest video effects box around, and pretty remarkable too.

The Voicetracker was a really neat concept, but failed to become a
marketable success.  It was a single small box which could accept an
audio input (from a microphone, say) and produce MIDI output which
tracked the pitch and timbre of the (monophonic) input.  The tracking
delay was on the order of a few milliseconds, and the pitch tracking
algorithm was very tight and accurate.  The MIDI output would drive
any synthesiser, thereby allowing automated doubling of vocal, sax,
flute, violin performances.

The CMI Series III was the culmination of the audio product line.  It
was based around the same dual M6809 CPU card that drove the Series II
but nearly all else changed.  The Floppy disk drives were supplemented
with hard disk controllers and a SCSI interface.  The system RAM
acquired rudimentary memory management and was expanded to 1MB.  The
entire audio architecture was revised and featured eight channel cards,
each of which could produce TWO monophonic voices.  There was up to
14MB of shared waveform RAM (not local to each channel card) with an
M68000 slave CPU to do numerical waveform crunching and sampling.
Sample rates of up to 200 kHz at 16-bit resolution were provided.
There was another slave M68000 with lots of timers and serial ports to
do real-time channel-card and MIDI i/o for the realtime music, sound
effects and digital tape sequencers.  The whole thing fitted into the
same small unit as the Series II, and sold for about $90k.
Unfortunately, this was the dinosaur which helped kill Fairlight.

The management of Fairlight Instruments made some pretty fundamental
mistakes in the introduction of the Series III.  In 1985, Series II
sales were going well, and R&D had been designing the Series III for
about half a year.  The decision had been made to keep the Series II's
8-bit data/16-bit address bus architecture for the CPU in order to
provide an `upgrade path' from Series II to half of a Series III.  The
hardware was being designed without the involvement of software people
and before any of the software had been designed.  Essential things
like indivisible test-and-set instructions were completely overlooked,
which was unfortunate in an architecture which included 13
microprocessor chips.

Then the Series III was prematurely announced to the marketplace, one
year before its scheduled completion.  Guess what happened to Series II
sales.  (Goodbye, cash flow.)  Suddenly the development schedule for
the Series III was of crucial importance, and in the rush decisions
were taken and dodgy software written which should never have seen the
light of day.  The Series III continued under the burden of its
involuted, undocumented and constantly fluctuating software (and
hardware) system until the end of its days.  Modules which should by
rights be simple to prototype and easy to complete took five to ten
times as long to write.  Complex ones could not even be attempted
except by the precious few who had helped build the systems in the
first place.  Learning curves lasted 6 months or more.  Engineers and
programmers became (almost) indispensable to the company.

This is not to say that the R&D staff were incompetent or
inexperienced.  In my opinion they were some of the most skillful and
inspired people I have met.  Most of them were involved in musical
activities outside work, and many had no academic qualifications.  On
reflection, working at Fairlight was a bit like spending time in a
happy but chaotic padded cell with tiny windows to the outside world.
Lots of fun but one developed a tendency to contemplate one's navel
overmuch.  Craziness was the norm.

In 1986 (my first year) things were getting tight, and Fairlight went
to the Management Investment Companies for help.  Fairlight Instruments
Pty Ltd had never shown any significant profits, and its shareholders
always voted to plough what profit there was back into R&D efforts.
Two companies, SAMIC and Western Pacific, invested large sums and
jointly appointed a director who was supposed to get things happening.
Things did pick up, another director was appointed and more money was
invested.  Fairlight USA, the American marketing and support company
continually gave trouble by not selling enough machines and consuming
too much of Fairlight's money.  200 Series IIIs were sold worldwide.
The "rich musician" market was almost saturated, and most sales were to
recording studios who wanted to make the CMI the centrepiece of their
service.

Fairlight identified the film and TV post-production audio market as
its target, and software development on the Series III swerved to
follow the new course.  Dialogue and sound FX editing packages were
designed, and Fairlight took the bold step into making pre-coding
functional and design specifications.  A new 68020-based waveform
processor card was designed and built to take the place of the old
68000 card.  Optical drives for archival were added and supported.  A
multi-track hard-disk digital recorder was designed.  Sales started
taking an upswing.

Then disaster struck.

In December 1988, Western Pacific reneged on an undertaking of their
board to give Fairlight Instruments a $2M payment which was needed to
hoist Fairlight into the next half-year, in which sales would
(supposedly) soar and profits would appear.  No amount of negotiation
seemed to help.  Fairlight was obliged (by law) to declare its
insolvency, and seemed likely to fall into receivership.  The bank
deliberately postponed the appointment of a receiver to give Fairlight
time to sort things out and get the required money.  Fairlight's
managing director spent the entire Christmas holiday period in
discussions with the bank and a crowd of interested potential
investors.  Western Pacific remained involved in negotiations seeking a
way in which they could continue to support Fairlight without exposing
themselves to what they evidently perceived as an unacceptable risk.
All the deals fell through.  The staff returned from holidays to see
the final deal with Western Pacific get rejected.

The bank appointed an agent to act as a mortgagee in possession.  The
first act of these agents was to stop salary and creditor payments and
investigate the possibilities of extending the overdraft or performing
liquidation.  They discovered that Fairlight had fewer liquidatable
assets than they had hoped, and so allowed the board of directors to
continue to seek outside investment.  Many of the staff took their
new-found position of zero income philosophically, and simply took
unpaid leave while Fairlight struggled to recover.  Many R&D staff
continued to work, with the immediate aim of consolidating all the work
to date into an instant software release.  When it became evident that
the no-salary position was going to be protracted, Fairlight discharged
its legal responsibility and retrenched the staff.

By this time all the locks had been changed, the factory had ground to
a halt and the office was being run on a shoestring.  Most staff sought
employment to keep themselves clothed and fed, and the remarkable
collection of expertise which had been the strength of Fairlight's
existence was disbanded.  Fairlight has now been liquidated.  Sadly, it
will probably be a long time before Australia hosts such a unique
concentration of skills.

This article should not be taken as an indictment of any of the parties
involved, but, if you like, take it as a general gripe about the status
of high-technology industry in this sunburned country of ours.
--
Andrew Bettison - Softway Pty Ltd
UUCP     uunet!softway.oz!andrewb   PHONE +61-2-698-2322
Internet andrewb@softway.oz.au      FAX   +61-2-699-9174
ACSnet   andrewb@softway.oz         SMAIL 79 Myrtle St, Chippendale NSW 2008
-------------------------------------------------------------------------