-1-

Should Musical Instruments be Dreams?†

Nicola Bernardini

Centro TEMPO REALE

Centro di Ricerca, Produzione e Didattica Musicale
Villa Strozzi
Via Pisana 77
50143 Firenze
Tel.++39(0)55/717270
FAX ++39(0)55/717712
E-mail: [email protected]

1. Introduction

Considering the pace at which technology evolves, 35 years is no small amount of time. 35 years is the age
of Computer Music, if Max Mathews’s early experiments in the IBM window in central Manhattan consti-
tute the starting point for all that has come since.
During these 35 years, we have dreamt plenty of different dreams. We have, over and over again, done and
re-done always the same job, a little better each time: get sound out of digital machines. By "we" I mean
all the musicians, scientists, composers, technicians, hardware specialists, music analysts, in brief, everyone
involved in threading music to computers.
In the beginning, this thread was really thin: only some truly stubborn scientist or some insane composer
with a strong faith would travel hundreds of miles to and fro with a tape full of output samples produced in
some university to have it played to the only available tube digital to analog converter somewhere in the
East Coast.
Today, the thread has become a glamourous avenue full of advertisements, marvelous shops, a few dark
alleys and some tramps sleeping here and there. Nowadays, the music market hardly considers traditional
instruments, scores, music paper, etc.: the real money-makers are small, self-demo-ing digital gizmos with
millions of preset sounds, huge hard disks, accelerator cards (for the bolder ones), the ever-growing class of
MIDI tools of some sort and the like.
That is certainly enough to bore me out of my own metaphors. At any rate, the question "Do we have what
we were looking for?" is still a hard one to answer.

† Written for the meeting to celebrate John Pierce´s 80th birthday, Northwestern University, Evanston, Il., 15-16th November
1990.

Nicola Bernardini — TR Ver. 1.3 Printed on 24/09/19

 -2-

2. Do we have what we were looking for?

The answer is, of course, yes and no.

2.1 Yes

Qualitatively speaking, we have everything we could possibly want for music processing, sound
acquisition, artificial performers, music analysis, sound synthesis and so on. Everything has carefully been
screened through extenuating triages. We have algorithms that will solve any problem quicker than ever.
Yes, it is true, this CPU is not fast enough, that hard-disk is not big enough, we need more RAM, more
DACs, more mice, bigger screens, etc. But we all know that this is a quantitative problem. The day we will
have all that speed, all that RAM, all that hard disk space and what not, we will want more and we will not
have answered the question yet.
That is in the nature of music and of musicians. Luciano Berio invited Peppino Di Giugno to IRCAM
because he needed a machine which could handle 1000 oscillators in real-time. Only a few years later, by
the time Di Giugno had the 4x machine, which was capable of handling 4096 oscillators and do much more
than just that, Berio was out of IRCAM, and everybody discovered, besides, that a large number of real-
time oscillators is not that significant musically and moreover it is humanly impossible to handle. But Di
Giugno was not out of his mind, and Berio was not just asking out of pure greediness: they were both,
along with many others, starting with John Pierce, Max Mathews, Jean-Claude Risset, John Chowning, and
so on, part of that long stream of visionaries who shared the joys and sorrows of musical research. Because
this is the name of the game here: musical research, the good old activity of Wagner and Heckel, Brahms
and Joachim, Debussy with the instruments of Adolphe Sax, Carl Maria Von Weber with the new clarinets,
and so on. Consider how many peculiar instruments music research has produced in centuries of harsh
work: the rankett, for example, or the sarrusophone, or the ophicleide. You can find hundreds of them.
That is only as far as musical instruments are concerned: you could even take a glimpse at musical forms,
stagings, or far-fetched compositional techniques (think, for example, of the cryptic counterpoint of the turn
of the century school of russian composers).
This is because musicians are not researchers: at the end of the day, their aims are different. Research may
fail but still be a valid experience. An artistic failure can hardly find a "raison d’être". Musicians search
for things that satisfy an immediate aesthetic requirement of some sort. When it comes to performing, any-
thing that fits this purpose will do: how far can that be from the abstract beauty and elegance of mathematic
formalizations of pure scientific research? On the other hand, the concert hall is not a lab (although it can
become one in the worst of cases).
So, we have everything: if we do not have it in real-time, we can get it in non real-time. If we do not have it
today, we will have it tomorrow. If we cannot afford it now, we will in the future.

Nicola Bernardini — TR Ver. 1.3 Printed on 24/09/19

 -3-

2.2 No

Nonetheless, we also know that we still need more tools. But what kind of tools do we need, since we have
everything? We still need at least two kind of tools:
1. Conceptual Tools
2. Interfaces
A computer is a very, very different instrument from traditional ones (including most of those used in
analog electronic music, which resembled traditional instruments enough). Maybe further thinking about
what it can represent for music and for musicians should be done. That’s what the conceptual tools are
about.
Chances are that when we have finished thinking (is it ever going to happen?) we will come out with some
ideas for man-machine interaction that are possibly not new but that have acquired the necessary theoretical
strength in the process. Then we will need to look into the interface business.

3. Conceptual Tools

A computer is a virtual environment which needs to be defined before any kind of functionality takes place,
let alone musical functions. This characteristic immediately sets it apart from traditional instruments. Each
one of these implicitly carries a set of historical, physical, cultural and linguistic elements which define its
musical space. In a computer, all these elements must be carefully organized in the compositional process.
This means subverting a great deal of the musical praxis of the past. By "musical space" I do not mean
simply choosing abstract linguistic and aesthetic structures and templates in which to set the creation of a
new work, as the traditional compositional practice would intend it, but the entire musical environment,
including multiple and compound environments, gestures, perception, etc.
This is not an easy task, even if, after all, virtuality is the ultimate necessity of music in the era of its
technisches reproduzierbarkeit. In fact, among other things, digital technologies allow complete seg-
mentation control over the acoustic and musical space: we can constitute "virtual music phonemes" with
which we can build new connotational dimensions of the musical message. This is achieved through
precise reproduction of dense messages (music) combined with the surgical tools of digital control of
sound. Music ceases then to be allographic and becomes an autographic form of art.1
To put it simply, what this means is that sound itself, (a specific instance of a musical performance)
becomes infinitely duplicatable and that, thus, the relationship between the listener and the work itself
changes radically: the listener is now confronted with something that resembles a photograph more than a
painting. In fact, virtuality itself allows complete control of connotation attributes: it is now possible to
hint, to lie, to fake, and even to create far-fetched metaphors with music. And furthermore still, it is
possible to lie in between these modes of communication moving continuously from one to the other, just
like we do when we speak (which is so hard to do when we write). In fact, this is the main issue: the com-
bination of reproduction and virtuality transforms music in a double layered grid language. The likeness
between its inner structure and functionality and that of spoken natural languages is now evident. Music,

1. for an exhaustive description of the terms allographic and autographic, cf. Nelson Goodman, Languages of Art, III,3 Bobbs-
Merrill 1968

Nicola Bernardini — TR Ver. 1.3 Printed on 24/09/19

 -4-

however, continues to preserve the auto-referential possibility of being ´understood´ just by listening to it.2

4. About this interface business

Given the concepts sketched above, it is easy to see that today´s composition requires a different approach
to the description of music and musical processes. There are at least two reasons for this:
1. scores are no longer a suitable vehicle of reproduction as they have been for centuries. There is not
only a graphic problem: in the past, the score was integrated by additional information such as
conductor’s indications composer’s annotations, stylistic elements, etc. Furthermore, it implies the
knowledge of instrument makers, the acoustical properties of concert halls, and so on. All this com-
bination of information is still available to us, of course; but digital tools allow us to integrate this
information in the compositional process: we must then be able to describe it in a deterministic
manner.
2. these different layers (timbre, interpretation, space location, etc.) become fundamental parts of the
composing process, requiring tools that, in the first place, allow to think about their organization and
structure.
The main problem, then, is a problem of description. As we are talking about computers, what this boils
down to, is user interface. We do not need endless power, we need endless interfaces.
Anybody even mildly involved in computers knows that interfaces are often ideologies, even faiths and reli-
gions at times. Everybody has his/her own preference, every computer book will more or less subtly induce
ideological concepts about the way a computer interacts with a human. I know many musicians who hate
to type but can spend hours and hours cutting and pasting, dragging and re-sizing, clicking and double
clicking. They look at a green phosphor screen like a princess would browse through Marx’s and Engels’s
Communist Manifesto. On the other hand, I also know the guru side of things, made of hundreds of in-line
commands with thousands of options all remebered by heart and the drawings that you only see when they
are printed (when it is most often too late, that is). Gurus despise windows and user friendliness a bit like
old south of Italy mammas despise young girls wearing miniskirts. The fact is, while technology changes
continuously interfaces tend to remain the same in order to be used by humans in a predictable way.
Therefore, I will try to focus as much as possible on the inner structure of interfaces and see how well they
suit possible descriptions of musical processes.
I think, for example, that not many will disagree with me on this (even though somebody will!): there´s no
sense in describing large-scale musical processes in a WYSIWYG (What You See Is What You Get) form −
which, in this case, would probably mean through a graphical sound editor. More meaningful descriptions
can be given through representational graphics or even through bare text. This is because music per se is
hardly WYSIWYG at all, it never was in the past and it never will be. How would you represent music
events, then? Traditional music notation is convenient to the player who has to read it and perform it. It is
far from convenient for the computer itself, which will engage in complicated graphic pattern matching
algorithms just to get a single note. In this case, text is much better: it is quicker to write on a typewriter
than to pick up notes with a mouse. It is extremely easy and convenient to find patterns (even complicated
ones) in a text file. PostScript™ demonstrates that text is a good language even for graphics description,
even though I would rather draw with a good graphical interface. The argument that typing text is more

2. An in-depth analysis of these mutations can be found in Nicola Bernardini, Semiotics and Computer Music Composition, in
Proceedings of the 1985 International Computer Music Conference, pp. 169-184, Computer Music Association San Francisco
1985.

Nicola Bernardini — TR Ver. 1.3 Printed on 24/09/19

 -5-

error-prone than picking up things with a mouse will crumble under my 180-character per minute blind typ-
ing vs. my total inability with small hand-held animals (a situation which, of course, can be reversed for
other people, thus proving the point).
This does not mean that text is good everywhere in music: try to edit a sound and you will discover that you
do not need the atomic discrete capacities of characters, and that a graphical interface with zooming capa-
bilities is extremely useful there. Also, if you need to build digital instruments by patching single functions
together a graphical abstraction like the connecting boxes "à la MAX"3 has proved to be a very powerful
tool. But again, it works well under specific circumstances (MIDI real-time interaction) which may not be
what you are looking for. If you need to do a basic musical gesture like laying out a scheduling of events,
then I personally find text to be the fastest, more convenient interface all bells and whistles excluded.4 Or if
you need to perform non real-time, hard-core processing of sound you may prefer data pipes better.
A last example.
MIDI became the de facto standard, in the music industry, for a musical communication protocol. Quanti-
tative limitations notwithstanding, this protocol is certainly a useful one if you have to describe events, it is
dreadful to describe processes. And music is obviously full of processes: try to describe a single-note
crescendo with MIDI and you will realize what I mean.

5. Conclusions

One of the important aspects of the computer as a tool for music making is its ability to keep its sonic out-
put in a field void of pre-conceived connotations and history.
To preserve machine trasparency and, thus, the virtual features of computers in music, though, we need to
have all interaction modalities available, possibly in the very same programs, and maybe with several com-
mon formats that will interconnect applications. We need program designers aware that, in music, no single
human interface is good for all applications and be able to pick that or those which will best fit the specific
purpose.
This is no easy task to achieve. Neither technically nor conceptually. Window-based applications are
structurally different from command-line based ones and they often imply a totally different attitude in pro-
gramming (e.g. single large applications vs. small interconnectible programs, etc.). And that is just the
start: large communication bandwith, speed of execution (an intrinsic one, this time), etc. are structural
obstacles to a good conceptualization of interface modalities.
At any rate, the solution of this problem is, in my opinion, the key question in future musical applications
of computers. We have the machines of our dreams, we need to dream about how to use them.

Acknowledgements

3. MAX is a graphical language devised by Miller Puckette at IRCAM

4. this is a good argument for discussion. Many people will work better with a graphical scheduler, as Morton Subotnick and
Mark Coniglio´s Interactor. I just happen to like writing better than drawing and to be more at home with it.

Nicola Bernardini — TR Ver. 1.3 Printed on 24/09/19

 -6-

I would like to thank Amnon Wolman for creating the opportunity of writing a contribution to the
celebration of the important life and work of John Pierce. In an interview he kindly gave me several years
ago, Max Mathews provided some interesting historical backgrounds on the beginnings of Computer
Music. Such thoughtful insights constitute the basis of many of my points of view. I am glad to have the
possibility to thank him publicly.
Francesca Ambrosi, Jane Fachiri Camilloni and Paul Roberts gave several fundamental suggestions during
the editing of this paper. They deserve my deep gratitude.

Firenze, August 31st 1990 Nicola Bernardini

Nicola Bernardini — TR Ver. 1.3 Printed on 24/09/19