Cube Octophony Daniel Laberge
     
   
        

 

 

 

 

 

 

 

 

 

What is octophony?

Octophony is an extension of stereophony and quadraphony.
Instead of two or four loudspeakers, it uses eight, placed at each corner of a cube and facing towards a center point.

Stereophony

Any stereo mixer has pan pots that let you place the sound on the left, on the right or at any point in between.
This point is often right in the center, and we are all accostumed to hearing an instrument in the middle when its volume is equal in both loudspeakers.

Strereo

The sound can be positioned anywhere between the two speakers,
on an unidimensional path, by proportionally adjusting its volume
for each channel.

 

Path for stereo sound

Path of a stereophonic signal

 

Pan button

Pan pots (panoramic potentiometers)
are used to place or move
individual instruments throughout the soundfield.

 

Quadraphony

In a quadraphonic system, also known as tetraphonic, you distribute the volume of a sound amongst four speakers to move it around on a plane field. This is usually accomplished with a joystick.

Quadraphony

The sound can be positioned anywhere between the four speakers,
on a two-dimensional square plane.
You hear it as being in the center when its volume
is equal in all four loudspeakers.
  

Quadraphonic path

Path of a quadraphonic signal

 

Joystick

A joystick is generally used to position or move the sound around.

 

Octophony

Add four more loudspeakers to a quadraphonic system, attach them to the ceiling, and you get octophony. This forms a cube figure with all the speakers facing a center point.

Octophony

When you distribute a sound source between the eight speakers,
you can place it anywhere in 3D space.
You hear it as being in the center when its volume
is equal in all eight loudspeakers.
 
  

Octophonic path

Path of an octophonic signal

 

Positioning sounds

Each individual instrument or voice can be placed anywhere along the soundfield. This is usually done to achieve a balanced and aesthetic distribution.
 

Typical stereo soundstage

Stereo soundstage

In stereo, each sound source can be positioned,
from left to right,
using the pan controls on a mixer board.
  

Similar positioning can also be achieved in quadraphony and octophony, but with the new dimensions added to the soundstage, moving the sound around then becomes more appealing.

Static or moving sounds

Sounds and instruments generally don't move; except for rare exceptions, they stay in the same place throughout a piece of music.
In stereo, this is understandable because moving sounds on a straight line, from left to right, is dull and quickly becomes annoying.

Stéréo soundstage

Stereo's limited soundstage

Motion on a quadraphonic soundstage

With four loudspeakers, a sound can pass right through the room and you can hear its position shifting across the square they form.

Quadraphonic motion

One sound moving on a quadraphonic flat soundstage,
Of course, several simultaneous sounds can each have their own path.

Motion inside an octophonic soundstage

Now, sounds can be moved across the tridimensional soundstage as if they were floating points in the air.

Octophonic motion

One sound moving freely across tridimensional space.
Several sounds and movements
are required for composition.

 

  

Mid-seventies

The idea of putting together an octophonic system came to me in 1975, when I was a composition student at the Faculté de musique de l'Université de Montréal.
I was working every week in the recording-composition studio, which was equipped with four large high-quality loudspeakers.
Putting four more speakers at ceiling level to achieve three-dimensional movement seemed to be a logical development.
I envisioned compositions where each voice or instrument would have its own movement in space.

Conceptual problems

I soon realized that many components of my octophonic system did not exist;
■ How could I assign a position to a sound in 3D space?
■ How could the eight outputs be calculated?
■ How could the sound movements be recorded?
■ Having one voice moving through space would be great, but I needed several simultaneous motions for my compositions.

A transducer

Even back in the 1970's, joysticks could be found on analog modular synthesizers.
They had to be connected to the synth's modules with patchcords.
I pictured that, to achieve 3D positioning, one could hook-up a joystick for x and y movements and use a knob to control the z axis; but operating this setup would be far from intuitive.
I needed a real three-dimensional transducer.
I envisioned a small cube, representing the 3D space of the room, and a wand with a trackable tip.

Finding a collaborator: Hubert Caron

The head of the Electro-acoustic department put me in touch with Hubert Caron, a young finishing student in engineering at the École Polytechnique de l'Université de Montréal.
Hubert was very bright and every facet of technology interested him.
My project aroused his curiosity and we soon began discussing the technical aspects.

Defining the project and looking for funds

This was a personal endeavour and it wasn't part of any academic program.
We decided to present a project to the Université de Montréal to officialize the status of our research, to get access to rooms and equipment and to get a little money for Hubert.
I would pay all the other project expenses on my student budget.

Getting to work

In order to present the following project, we had to practically design the whole system and test some parts of it in advance.
We started working vigorously on octophony, whatever the outcome.

 

  

This chapter is in french. English text resumes after it.

Étude du déplacement
tridimensionnel du son

Projet présenté par: Daniel Laberge
Avec la collaboration de: Hubert Caron
Montréal, le 10 octobre 1977

Introduction

Ce rapport est un exposé de mon projet de recherche et de son cheminement.
Il est divisé en six parties:

1■ Présentation et description du projet
2■ Expériences antérieures
3■ Étapes du projet
4■ Description technique des appareils à réaliser
          A■ Contrôle de balance
                    1■ Transducteur
                    2■ Contrôleur de gain
          B■ Encodeur, décodeur
5■ Expérimentation et composition
6■ Cheminement et allocations

1■ Présentation et description du projet

La stéréophonie à deux canaux, puis la tétraphonie, ont tour-à-tour tenté d'améliorer la qualité de la reproduction de la musique enregistrée en y apportant une nouvelle variable; la dimension spatiale.
Sur chaque mélangeur dit «stéréo», on trouve des contrôles de panorama qui nous permettent de déplacer le son de gauche à droite, sur une ligne droite; donc en une dimension.

Stéréo

Ce système implique le concours de deux haut-parleurs et d'un amplificateur jumelé. La fabrication d'un contrôle de panorama est techniquement assez simple.
 
La venue sur le marché de chaines d'appareils dits «tétraphoniques» ou «quadriphoniques» a permis le déplacement du son sur un plan, donc à deux dimensions.

Quadraphonie

L'installation nécessite quatre haut-parleurs et un amplificateur à quatre entrées et sorties indépendantes. L'industrie offre un contrôle de balance type «joystick» qui consiste en un bâton vertical qu'on peut déplacer de tout côté et qui agit de la même manière sur le son.
 
Mon projet est l'extension finale des deux principes mentionnés plus haut en y ajoutant la dimension manquante.
Le théâtre de l'action est donc un cube.

Cube

J'aurai besoin de huit haut-parleurs et de quatre amplificateurs stéréophoniques. Mais l'instrument qui permettrait de déplacer le son librement à l'intérieur d'un cube (pièce) n'est pas encore inventé.

2■ Expériences antérieures

Des expériences semblables ont déjà été tentées par des compositeurs comme Stockhausen et Xenakis. Leurs recherches avaient cependant des bases et des résultats différents des miennes.
À la présentation d'une oeuvre de Stockhausen, des haut-parleurs tapissaient le plafond et les murs d'une pièce, le compositeur manipulait une lampe de poche à l'intérieur d'une maquette de la pièce contenant des cellules photo-électriques correspondant à chacun des haut-parleurs. Les cellules affectées par le faisceau lumineux mettaient en action les haut-parleurs. L'auditeur pouvait ainsi entendre le son se déplacer sur les murs de la pièce.
Il ne s'agissait toutefois pas d'un contrôle de balance qui aurait pu permettre à des haut-parleurs opposés d'entrer en action et de créer l'impression que le son se déplace dans l'espace de la salle et non pas sur les parois.

3■ Étapes du projet

Le projet se divise en trois phases:
 
1■ Construction des appareils:
          A■ Contrôle de balance
                    1■ Transducteur
                    2■ Contrôleur de gain
          B■ Encodeur-décodeur (multiplexeur)
  
2■ Expérimentation
3■ Composition d'une pièce musicale explorant les possibilités de cette nouvelle scène d'action.

La faculté de musique consent à me fournir les locaux et appareils supplémentaires, nécessaires à l'exécution des deux dernières phases de ce projet.

4■ Description technique des appareils à réaliser

La partie technique du projet comprend la réalisation d'un contrôle de balance tridimensionnel agissant sur un système de huit haut-parleurs ainsi qu'un système d'encodage et de décodage permettant l'enregistrement et la lecture sur bande magnétique de l'information de balance.
Les huit haut-parleurs du système de reproduction sont placés à chaque sommet du parallélépipède rectangle formé par la salle de concert (voir figure 1). On peut donc, en variant l'intensité sonore relative (balance) émise par les haut-parleurs, faire changer la provenance mutuelle du son à l'intérieur de la salle. Il est à noter ici que, bien que les relations de phase entre les signaux émis par les différents haut-parleurs aient une influence indéniable sur la perception spatiale d'un son, je ne m'attacherai ici qu'à l'aspect amplitude (volume) de ceux-ci.

A■ CONTRÔLE DE BALANCE
1■ Transducteur
Le transducteur lui-même sera de nature opto-électronique. Il convertira les trois coordonnées de position d'une source lumineuse se déplaçant à l'intérieur d'un cube (représentant la salle d'audition) en trois signaux électriques respectivement proportionnels (voir figure 2).
La source lumineuse pourra être montée sur une tige manipulée par le compositeur lui-même. L'analogie entre la salle d'audition et le volume cubique du transducteur en rend la manipulation évidente et ne requiert aucune connaissance technique préalable. Il suffit, par exemple, de monter la source lumineuse pour privilégier les haut-parleurs situés au plafond ou de la baisser pour que ceux situés au sol jouent plus fort.

 

Figure 1

Disposition

Disposition des haut-parleurs aux huit coins de la salle.
Chaque haut-parleur est orienté de façon à viser un point imaginaire
au centre de la pièce.

 

Figure 2

Transducteur

Le transducteur transforme les trois coordonnées de position (x, y, z)
de la source lumineuse en trois signaux électriques
respectivement proportionnels.

 

Figure 3

x/x' ratio

Dans ce montage, le rapport x / x' est relativement grand
de manière à permettre au capteur de réagir de façon prépondérante aux déplacements axiaux de la source lumineuse
et le moins possible aux déplacements latéraux.
Pour diminuer encore davantage la réponse latérale,
un filtre, dont la transmittance varie radialement,
est mis devant le capteur optique pour modifier sa réponse angulaire.
La réponse angulaire est donnée ci-dessous.

 

Réponse angulaire

Graphique de la réponse angulaire du capteur optique

 

 

Figure 4
Contrôleur de gain

Tableau

Le tableau ci-dessus donne le gain de chaque haut-parleur
ainsi que sa valeur algébrique.

 

Numérotation des haut-parleurs

Cette figure indique la convention
de numérotation des haut-parleurs utilisée.


On fixe la valeur maximum de tout voltage de contrôle à 1volt
ce qui implique que;

Le diagramme suivant donne le circuit
nécessaire au calcul des huit gains

Diagramme
Légende

Le contrôleur de gain produit, en effectuant les calculs analogiques appropriés sur Vx, Vy, et Vz, les huit signaux de contrôle (1 à 8)
qui règleront le gain de chaque amplificateur.

  
  

Figure 5

Montage des V.C.A.

Dans ce montage, chacun des huit canaux audio est généré
par trois V.C.A. ou «voltage controled amplifier».
Chacun des V.C.A. contrôle ou multiplie le signal audio d'entrée
pour chacun des trois voltages de contrôle
formula ou leur complément Formula.

 

Si les voltages formula varient entre 1volt et Vr, les voltages complémentaires Formula sont respectivement égaux à;
                                 Vr - Vx,   Vr - Vy
   et   Vr - Vz .
Si le volume sonore dans un haut-parleur est proportionnel à;
Formula, le haut-parleur qui lui est diamétralement opposé produira un volume sonore proportionnel à Formula.

Ce système est composé de 24 V.C.A. et de trois soustracteurs
pour produire les voltages complémentaires.
De plus, l'absence de toute contrainte mécanique (bouton à tourner ou curseur à glisser) permet au manipulateur de produire de très brusques et rapides variations dans la balance sonore.
Différentes possibilités existent pour repérer la source lumineuse à l'intérieur du cube; la seule contrainte importante étant de conserver à tout instant l'indépendance dans la perception des trois coordonnées. La figure 3 montre un montage possible.

2■ Contrôle de gain

Pour que l'information (3 signaux) produite par le transducteur puisse contrôler le volume restitué par les huit haut-parleurs, il faut que ceux-ci soient traités par un contrôle électronique de gain.
Pour minimiser le nombre de composantes requises et le bruit produit inévitablement par les amplificateurs, la structure en arbre de la figure 4 a été préférée à celle de la figure 5. Dans cette dernière, le signal audio lui-même est traité accessoirement dans trois V.C.A. (voltage controled amplifier) montés en série et contrôlés par les trois coordonnées ou leur complément. Celui de la figure 4 n'agit d'abord que sur les signaux de position (qui se distinguent nettement du bruit et peuvent ainsi être aisément filtrés) pour produire huit signaux de contrôle de gain agissant sur huit autres V.C.A. qui eux traitent l'information musicale.
Un autre aspect intéressant de ce contrôle de gain est que les signaux d'entrée peuvent provenir d'une autre source que du transducteur déjà décrit. Des effets analogues aux figures Lissajou, produites sur un écran d'oscilloscope, peuvent être obtenus sur le plan sonore.

B■ Multiplexeur

On pourrait bien sûr enregistrer sur des pistes séparées les huit canaux qui sont reproduits par les huit haut-parleurs, mais il y aura redondance de l'information. C'est-à-dire que quatre informations seulement sont nécessaires et suffisantes pour décrire parfaitement à chaque instant l'état du système:
          ■ Les trois coordonnées rectangulaires de position (x, y, z ) ou de balance.
          ■ L'information sonore elle-même (musique).
Les trois premières informations ne contenant que des fréquences très basses (subsoniques), de 0 à 10 Hz environ, peuvent être facilement multiplexées et toutes enregistrées sur la même piste magnétique, alors que l'information musicale occupe une piste séparée.
Dans le système proposé, chaque signal de balance module en amplitude une porteuse de fréquence fixe avant d'être mixée et enregistrée. À la lecture, une série de filtres passe-bande permettent de récupérer indépendamment chaque porteuse. Un démodulateur en extrait le signal initial. C'est finalement ce dernier qui agit sur le contrôle de gain.

5■ Expérimentation et composition

Dès que le transducteur et le contrôleur de gain seront terminés, il sera possible de faire les expériences préliminaires qui démontreront l'étendue réelle de la perception du déplacement sonore.
Ces essais, s'ils sont concluants, serviront de base à la composition d'une pièce de musique électronique à trois ou quatre pistes. Chacune de ces pistes aura son propre déplacement.

6■ Cheminement et allocations

Chacun des deux volets de la fabrication des appareils peut être réalisé complètement en cent (100) heures pour un total global de deux-cents (200) heures (des heures supplémentaires pouvant être ajoutées au besoin).
Le projet serait échelonné sur l'année universitaire 1977-78. La première partie peut être complétée pour décembre 1977, et la seconde pour avril 1978.
Monsieur Hubert Caron dispose d'équipement technique personnel qu'il utilisera pendant ce projet.
Je demande les sommes suivantes pour la réalisation du projet: mille ($1,000) dollars pour chaque partie, soit un total de deux mille ($2,000) dollars.

 

  

The project takes off

Our demand was accepted, so we went straight to work.
Since most of the conception of the different elements of the system was already done, we could start building them.

Doing everything ourselves

Hubert was skillful, dextrous and clever; few technical hurdles could stop him.
We designed and constructed everything from bits and pieces.
We worked every Sunday, from morning to evening. During each session, we made a list of what we needed for the next time; so I would go out on weekdays to buy the parts and material.

The tridimensional transducer

Tridimensionanal transducer

Illustration of the tridimensional optic transducer and illuminated wand. Towers, holding the opto-electronic captors,
are mounted on three faces of the 18 inch cube.
 

The structure of the transducer is made entirely of aluminum.
Each captor «sees» its own axis, respectively x, y and z, but radial optical filters were placed in front of each photo-cell to eliminate crosstalk from the other axes.
We had to calculate the curve for these filters, experiment with the results and make them ourselves.

Operation

The transducer is conceived to be operated inside a larger cube formed by eight loudspeakers.

Operation of transducer

The sound inside the room follows the position
of the light inside the cube.
 

More equipment was typically present or accessible in the room along with the composer:

  ■ The main unit of the octophonic system (see below)
■ The tridimensional figure generator (see below)
■ A modular analog synthesizer with keyboard
■ The remote control for an eight track tape recorder
■ A mixing board

The process, step by step

1■    Record an audio performance on track one Track allocation

 

2■    Record the spatial movements for it on track five Track allocation

Steps 1 and 2 could also be executed simultaneously.

3■    Record an audio performance on track two Track allocation

 

4■    Record the spatial movements for it on track six Track allocation

 

5■    Record an audio performance on track three Track allocation

 

6■    Record the spatial movements for it on track seven Track allocation

 

7■    Record an audio performance on track four Track allocation

 

8■    Record the spatial movements for it on track eight Track allocation

The main unit

The main unit housed the circuit boards for the various modules needed to move the sound around.
They included:

  MODULES   TASKS
The signal shaping module   It receives the 3 coordinate voltages from the transducer. It then shapes and amplifies them so they can be used as control voltages.
       
The FM encoder module   It converts the 3 position control voltages so they are recordable on one audio track.
       
Four FM decoder modules   They convert the position signals recorded on tape back to control voltages.
       
Four gain controller modules   They take the three x, y, and z coordinates and compute how the volume should be distributed through the eight speakers in order to place the sound in tridimensional space.
       
The output module   Where the signal for each speaker is accessible.

 

The cabinet, containing the parallel circuit boards with bus connectors at the back, was constructed from scratch.
Each board was designed, pre-tested, drawn, photographed, etched, mounted with soldered electronic parts and tested.

The 3D figure generator

The tridimensional transducer, with its wand, worked perfectly for complex spatial motions that follow directly the music. A bit like a conductor.
But sometimes, I felt the need to do a relatively simple figure, such as a circle, line, square, ellipse, ..., but it had to be perfect and slow.
Many ideas and details for the figure generator came later in the project, so it was done last. It was mostly digital circuitry, because of the need for stability and phase locking.
 
Many people are familiar with images likes these:

Lissajou figure   Lissajou figure

They are Lissajou figures. They are obtained when the frequencies of two plotted waves are entire multiples or fractions of one another. Changing their relative phase transforms the patterns.

Basic Lissajou figures

Lissajou figures

These figures are generated using two sine waves. The figure generator could also produce triangular and square waves; so square, rectangular and hybrid figures were also obtainable.

Furthermore, the figures above are only two-dimensional, while the generator added the third axis to produce much more complex trajectories.

Lissajou spiral

Perfect spirals could easily be achieved with the figure generator.
They could be oriented in any direction and their speed of execution could be set or controlled manually in real-time.

The 3D figure generator was contained inside a handheld, six inch, metal box and connected to the main unit with a long cord, permitting its use at any point inside the cube made by the loudspeakers.

 

  

First octophonic tests

In the summer of 1979, the system was ready for its first «in situ» tests; in a large room, with four speakers attached to the ceiling and four more at floor level.
I had been using the device for some time in quadraphony.

The setup

Since occupancy was low during the summer period, I reserved the main concert hall of the faculty for a block of consecutive days.
The room was flat, with no fixed seats, so the entire space could be freed for our tests.
I borrowed eight high quality loudspeakers from the Electro-acoustic department and we fastened four of them, at the correct angle, to the ceiling of the room (take note of the ladder at the back of the following photo) with the other four speakers right below them in order to produce an eleven foot cube.
 Four high-quality stereo amplifiers linked our system to the loudspeakers.
 

The octophonic transducer in action

Photo of 3D optic transducer

Photo of the three-dimensional opto-electronic transducer.
You can see the cube, made of aluminum tubes, with towers extending in the three x, y, z directions. The opto-electronic sensors are placed at the summit of each tower. Any movement by the wand, with its luminous tip, is transmitted to the main unit as three coordinates.
Université de Montréal, summer of 1979

Calibrating and testing

Since each module in our system offered extensive calibrating capabilities, we worked several days at making sure every voltage corresponded to Hubert's calculations.
 

Testing the octophonic system

Photo of optic transducer during tests

3D opto-electronic transducer seen from the back.
The main unit, with its vertical circuit boards, is directly below it,
next to four stereo amplifiers.
This equipment is situated inside an 11 foot cube
formed by eight loudspeakers.
Université de Montréal, summer of 1979

The system passes the test

An octophonic system has to be symmetrical on all three axes.
A figure generated in one part of the cube has to be replicable in any other one.
We found that the system performed in the way that was expected, both in electrical tests and in listening tests (described below).

Someone stole
the four ceiling loudspeakers

After this summer test period was over, we cleaned the room of the equipment we had brought in, but decided to leave the four ceiling loudspeakers in place so we could keep on testing during the fall session.
Attaching and orienting them had taken some time and their presence wasn't a disturbance.
I was informed, at the start of the academic year, that the speakers had been stolen.
This event had a disastrous effect on the project.

 

 

Listening tests

How octophony sounds

Hearing the sound moving freely in 3D space is both exciting and intriguing, the effect is new to our ears.
It makes you focus your attention on what is happening behind, above, below and all around you.
Of course, the perception we have of a sound differs greatly depending on its provenance in relation with our outer ear.
So, even if the octophonic system is entirely symmetrical, the human ears have evolved to privilege sounds coming from the front and at their height.

The perception of motionless
and moving sounds

Studying the «movement» of sound in tridimensional space was the title of the whole project, not its fixed positioning.
This is probably because, long before building anything, I was convinced that:

Movement improves our perception
of a sound's position in space

Static sounds, those that stay in the same place for an extended time, are much harder to localize in space than those that move.
It is as if the positional change alerts your brain of the sound's presence and it then keeps track of its trajectory.
Our listening tests confirmed this theory.

Speed limit

We soon found out that the brain could only follow a sound's movement up to a certain speed.
Passed that speed, the brain gives up and considers the motion as a swirling effect.
For example, if the sound follows a simple circular pattern around the room, its position can be easily followed when each rotation is done in two seconds. The slower you go the better the perception gets, up to a limit.
At faster speeds, above one rotation per second, it becomes blurry and punctual detection ceases at around half a second.
Even if localization is lost at higher frequencies, the motion gives the sound a new and unique dimension and has proven to be, to me, an interesting compositional tool.

Pre-recorded or live music

While composing with the system, I found that performing the music live at the same time I was directing it in space yielded superior results, when compared to using pre-recorded music.
Obviously, giving the spatial dimension equal importance with sound creation helps localization, mostly when they are generated simultaneously.

High and low pitched sounds

We were well informed, before starting this project, that the perception of a sound's localisation, in humans, diminishes as its frequency goes down.
This is why only one central loudspeaker is used for bass in several stereo configurations.
We also toyed with the idea of excluding low frequencies from our system, but we decided otherwise.
This ended up being a wise decision because I found that I could follow some low sounds around quite well.
With motion, especially when a musical pattern is present, the position of low pitched sounds becomes more perceptible.
This phenomenon can be further enhanced by using localisation helpers, as described below.

Attack and sustain phases

It is important to point out that the localisation of a sound in space is best performed during its initial onset, or attack period.
The first portion of each musical note is very identifiable because its volume is at its loudest and that several of its parameters vary very quickly.
Simply repeating a sound, even when changing its pitch, will retrigger its positioning by the brain.
Later on, when the sound is sustaining, localisation is diminished and can be lost totally if no change in timbre is detected.
Therefore, it is best for the composer, when using long notes, to keep on changing the sound's parameters while they are sustained.

Localization helpers:
the sound-position interactions

I also immediately noticed that localization was greatly enhanced when one of the music's parameters followed the positional values for one of the three axes.
The simplest example I can give is having the frequency of the sound go down as the wand goes down, and up when it goes up.

The simplest sound-position interaction:
the  z  coordinate controls pitch

High-Low

When the sound goes up and down along with its position in space, localization perception is improved to a great extent.
This relationship seems to be obvious to the human brain.

Inverting or modifying the orientation doesn't cancel this predisposition, but alters it.

Inverted Highs and lows

Inverted and modified orientations

Other sound-position interactions

Once this concept is understood, it can be applied to any controllable sound parameter.

Sound-position interactions
z coordinate
controlling
the filter's
cutoff frequency
z coordinate
controlling the
reverb send
z coordinate
controlling the
vibrato's (LFO)
speed

Since the x, y and z coordinates for each voice were readily available as control voltages, all I had to do was to plug them in the modulation input of any synthesizer module and it would respond to spatial positioning.
All these «sound-position» to «sound-parameter» relationships are helpful in localizing it.

3D sound

As a result, by hooking up the three spatial coordinates to a synthesizer's various modulation inputs, I could set up sounds that were different in every part of the space.

3D sound

The characteristics of a sound traveling along this figure
change depending on its position in space.
In fact, the sound would be different at any point inside the cube.

In some configurations, the sound can become muffled, muted and can even disappear completely in some parts of the cubic space.

Silence in part of the cube

The listener's brain fills the gap

While this phenomenon would prove unacceptable for most types of music, in the case of 3D motion, it almost seems natural.
The brain will make up for the missing section, as long as it isn't too long.

Two, three and four
simultaneous sound movements

As expected, having two or more sounds, each one with its own tridimensional spatial travel, at the same time inside a cube makes their localization more difficult.
How well you will perceive their location depends on how different the sounds, their positions and their motions are one from the other.
Slower moving ones are better identified.
This being said, aesthetically interesting composition only begins, for me, when two or more voices are used. The listener's brain may not be able to follow their whereabouts, but the air of the room moves and pulsates along with them.

Mirrored 3D motions

Having four sounds doing completely different movements is interesting, but linking the spatial paths of some voices together heightens their perception and impact.
Once one spatial trajectory is recorded on tape, its three x, y and z coordinates can be resent during playback to a second sound, but this time they can be inverted, modified or rerouted.

Mirrored motions

Recorded trajectory
for sound 1
  The spatial coordinates
of sound 1 are reused
for sound 2,
but one of them
is inverted

Mirroring offers a way to synchronize, combine and transform spatial figures; creating multi-voice 3D sound patterns.

The «center proximity» coordinate

Not all points are equal inside an octophonic cube.
The ideal location to place your head is right in the center.
Other positions may also be adequate, but you get an altered version of the music and figures.
When a listener's head is in the center of the cube, all the sound's movements and figures are perceived from that vantage point.
Some sounds are close while others are distant.
There are several distinctions between sounds that are near you and those farther away. For example; their phase (in relation with your ears) changes, their reverberation characteristics differ, their timbre is modified, ...

center proximity effect

The value of the «center proximity» coordinate
varies as the sound moves away from the center point.

The «center proximity» coordinate can be used by a composer in several ways to enhance the distinction between sounds that are moving towards and those moving away from the listener.

Synchronizing music and figures

We had included an external clock input on the 3D figure generator so it could be controled by other sources such as sequencers and synthesizers.
This has proved to be an essential feature.
It permits the interlocking of individual notes with their position in space.

Synchronisation

One bar musical pattern synchronized with a circle figure.
Every time it repeats, the notes stay in the same place.
The length of the sequence, its musical content and the figure it follows can all be chosen by the composer.

Both pattern identification and spatial position recognition are helped by synchronisation.
Even with non-repetitive musical motives, it allows for each bar to start at the same location and for each beat to have its own region in space.

Composing with the system

As explained below, I used the system mostly in quadraphony.
I composed with it for several hundred hours, during a two year period, in a quadraphonic environment, with a four track tape recorder at the Faculté de musique.
I also had access to a large modular analog synthesizer with a keyboard and a professional mixer.
I was composing a 20 minute piece of music dedicated to exploiting sound movement.

Musical style

As a composer, I have never let myself be restrained to a specific style; I feel free to go anywhere I fancy.
For this piece, I chose to work with electronic sounds rather than using acoustic instruments because they permitted a better interaction between the sound and its position in space.
I ended up doing a type of music I had never imagined before.
Since every aesthetic musical decision I made took into account the beauty of the spatial figure that accompanied it, the result was like a musical sculpture.

Sound choreography

Each voice is comparable to dancer on stage; with several of them, you can create choreographies.
But with tridimensional motion, each musical track becomes more like an individual plane during an air show.

Sound show

Sound show

The people watching these shows enjoy the patterns and tightly choreographed motions the planes make.
However, if the planes didn't leave a smoke trail, the figures they produce would be much harder to perceive.
I also believe that our brain is more inclined towards visual pattern recognition than sound motion ones.
Finally, in octophony, since the head of the listener is situated in the center of the action, the figures are perceived as if he was immersed in them.

Composing for moving sound

I really enjoyed the excitement motion brings to even the simplest sound.
Filling out the empty space became more of a preoccupation.
Since the sound surrounded me, I felt part of it.
I ended up doing about twenty short «musical animations».

The aesthetics of moving sound

As I composed these pieces, I became aware that the music I was making would not be appreciated by my teachers.
I was facing new musical surroundings that forced me to innovate and to expand my definition of music.
A simple sound, moving around in space while its timbral characteristics are changing, may give interesting aesthetical results.
I was designing multi-voiced choreographies and sound sculptures that had little in common with the classical, popular or contemporary music languages
What I was creating could not be written down or analysed with traditional methods.
Most musical genres, when recorded in stereo, can still be enjoyed on a monophonic system.
This was not the case of what I was doing; it had to be experienced in the proper environment for it to make any sense.

Technical restrictions

Four loudspeakers instead of eight ...
A four track tape recorder instead of an eight track ...
I know that the Faculté de musique de l'Université de Montréal is now wonderfully equipped for recording, but that wasn't the case in the 1970's.
In 1979, I wrote the following letter to the dean of the faculty, asking that they buy an eight track tape recorder.

 

Le 15 août 1979

Monsieur le doyen de la Faculté de musique
2375 côte Sainte-Catherine
Montréal, Qc.

Monsieur le doyen,

J'ai déjà, il y a près de deux ans, fait appel à votre aide, au sujet du projet «Étude du déplacement tridimensionnel du son». Il s'agissait, à l'époque, de trouver des fonds pour rémunérer M. Hubert Caron, ingénieur, qui travaille avec moi. Je suis étudiant en rédaction d'une maîtrise en composition.

Le projet comprend trois phases:
1- Construction des appareils.
2- Étude du déplacement tridimensionnel du son.
3- Composition d'une pièce en «octophonie».

La première phase, qui nous aura demandé plus de 2,000 heures de travail, sera terminée d'ici deux mois.

Le but de cette lettre concerne les deux phases subséquentes. En effet, dans notre système, les trois signaux de position (x, y, z) sont traités indépendamment du signal audio et «encodés» (multiplexés) sur une piste parallèle à la trame musicale. Voici un diagramme indiquant la répartition des pistes d'une pièce à quatre voix distribuées dans l'espace.

           Signal audio; 1, 2, 3, 4

Coordonnées de position: 1, 2, 3, 4

Évidemment, ce système exige l'utilisation d'un magnétophone ayant huit pistes au minimum, et la faculté de musique de l'Université de Montréal n'en possède pas encore.

Je demande donc l'achat d'un tel appareil qui serait aussi très utile au travail de composition électro-acoustique. Connaissant les difficultés que cet achat pourrait occasionner à la faculté, je suggère l'achat d'une huit pistes semi-professionnelle de marque Tascam et de modèle 80-8 dont le prix institutionnel se situe aux alentours de $4000.

Je me tiens à votre entière disponibilité au cas où les renseignements fournis dans cette lettre s'avéreraient insuffisants.

Veuillez agréer, monsieur, l'expression de mes sentiments les meilleurs.

 

Daniel Laberge
27xx place Darlington, apt. 24
Montréal, Qc.
H3S 1L4

 

Work summary

The project lasted over three years .
Instead of the 200 hours we thought it would take, we invested thousands.
Except for the initial $2,000, Hubert never was paid afterwards.
Though the expenses weren't considerable, they ate up my budget as a student.
We did it because we enjoyed building and inventing things.
Octophony was a challenge, so we put the time and efforts needed to achieve it.

 

  

The 1982 International Conference
on Acoustics, Speech and
Signal Processing in Paris

I don't recall who suggested that we submit our project for the ICASSP conference.
It was, for us, a way to get a little recognition for our efforts.
In order to get accepted, we had to write a paper on octophony, with strict guidelines, and submit it to their jury.

Our paper is accepted

Our article was selected and we were scheduled to give a conference in Paris, in the morning of the 5th of May.

ICASSP 82 program

The ICASSP 82 international conference
advance program cover
 
 

Program inside

 

Detail
Detail

ICASSP 82 conference program details

Funding again

We expected the university would help with our travel expenses, but we got a negative.
We were only asking for about $1,000 each, for airfare and lodging.
I decided to write to the Ministère des Affaires Culturelles du Gouvernement du Québec.

 

  

OCTOPHONY

Hubert Caron, Daniel Laberge

Faculté de musique, Université de Montréal
Montréal, Québec

ABSTRACT

This paper describes a study on three-dimensional sound panning used in electro-acoustical music. A complete electronic three-dimensional panning system has been developed, evaluated and successfully used in a live performance.

The main features of this system are: real-time operation, human-engineered opto-electronic controller, compatibility with standard studio recording equipment and ease of interfacing.

Psycho-acoustical tests have been conducted and prove that the system is a valuable tool for the electro-acoustic music composer.

SYSTEM OVERVIEW

«Octophony» can be seen as an extension of two-dimensional panning, known as tetraphony or quadraphony, used in electro-acoustical music.
Tetraphony uses four speakers to create a bidimensionai scope (X and Y) so the sound can be moved on a plane surface.
«Octophony» is the final extension to these systems and adds four more speakers (at the ceiling). This cubic configuration of speakers permits the use of the «Z» axis (along with X and Y) to achieve controlled motion of sound in three-dimensional space.

In our view, the system had to meet the following five requirements:

  ■ The ability to independently control the three dimensional panning of four separate musical tracks or voices.
■ Give the user immediate response to his command.
■ Design a sound balance controller allowing him to easily draw any three-dimensional figure he can imagine.
■ Provide a compact and handy way of recording music and balance information on a standard studio tape recorder.
■ Maintain system compatibility with other existing equipment such as music synthesizers, sequencers and other audio processing devices.

Modular construction has been used extensively in the design of all the system's electronic circuitry. Figure 1 shows some of the more important modules and how they are interconnected.

The leftmost modules are balance or panning control voltage sources. These are: the opto-electronic controller which is a manual three-dimensional balance control, a three-dimensional figure generator that is used to create automatic and repetitive figures and finally, an FM decoder module which retrieves control voltages already recorded on tape.

The other modules are used to diffuse the musical or audio input into the eight-speaker system, as a function of the X, Y and Z control voltages. An FM encoder module is also provided to record control voltages on tape. All of these modules will be covered in more detail in the following sections.
 
 

Figure 1

System block diagram

SYSTEM BLOCK DIAGRAM

 

GAIN CONTROLLER

The heart of the system is the gain controller module (there are four of these to control four separate voices) which is used to control the
respective output of each of the eight speakers according to the instantaneous values of three control voltages representing respectively the X, Y and Z sound balance.

This is done by first computing eight control voltages proportional to the respective power output of each speaker and then applying each control voltage on a voltage controlled amplifier through which the audio signal is processed before being sent to a power amplifier and speaker. Of course, VCA's must exhibit a linear power to control voltage characteristic in order to maintain constant total power output in the system and linearity of balance control.

To maintain compatibility with existing equipment, control voltages are unipolar 0 to I0 volts, 10Hz bandwidth signals. The eight VCA's control voltages will then be obtained from the normalized product of the three balance control voltages or their center-symmetricals:

Gain calculations

where center-symmetrical of

Formula


In the present implementation of the gain controller, analog computation has been used to obtain V1 through V8 (some algebraic simplification has been done to reduce the number of analog multipliers needed).

THREE-DIMENSIONAL
OPTO-ELECTRONIC TRANSDUCER

The purpose of this transducer is to give the operator a real-time manual control over the three-dimensional balance of a sound source.

The transducer appears as an 18 in. cubic aluminum structure which happens to be a scaled down model of the listening room. The three-dimensional position of a light point-source (held by the operator) relative to the center of the cube is converted into three control voltages which are in turn fed to the gain controller described above. With such an arrangement, the resulting sound balance will follow any movement of the light point-source imparted by the operator.

The position to voltage conversions are handled by three phototransistor modules respectively sensitive to X, Y and Z coordinates. In order for each module to respond only to one coordinate, a filter with radially varying transmittance had to be placed in front of each phototransistor.

THREE-DIMENSIONAL
FIGURE GENERATOR

This figure generator offers an additional way to control the three-dimensional sound balance. It is composed of three low frequency waveform generators having sine, triangular and square wave output.

If the frequency doubling switch provided for each generator is not used, the three run at the same frequency which is adjustable by a single knob. One generator acts as a «master» unit to which the other two «slave» generators are phase locked. The phase shifts between the two slaves and the master are separately and continuously adjustable.

Any three of the nine generated signals (3 waveforms from 3 generators) can be connected via a patch panel to three output amplifiers offering gain and offset control.

Very low frequency phase stability has been obtained through the use of digital circuitry which also offers the possibility of synchronizing the generators with other sources like music synthesizers and sequencers. A TTL compatible «external clock» jack has been provided for this purpose.

Through the use of this figure generator, sound balance figures similar to the well-known Lissajou figures can be obtained.

FM ENCODER DECODER

Once the audio signal (musical track) has been processed through the gain controller module, the resulting eight audio signals could be recorded on any conventional multi-track tape recorder. However, for reasons given below, it has been chosen to record separately the unprocessed audio signal and the balance information comprising the three control voltages applied on the gain controller module, these being FM encoded due to their very low frequency content (0 to 10Hz) and multiplexed on a single track.

Among the many advantages of doing so, let's just mention:

  ■ the compactness of the resulting recording: one track for the musical signal, plus one more for the multiplexed FM encoded balance information, as opposed to the eight tracks
needed for direct audio recording.
■ the ability to retrieve, reprocess or modify the balance information while leaving the musical track intact. This feature could be used to create «daughter» balance tracks that are processed copies of a previously recorded «mother» balance track. As an example, synchronized, symmetrical notation figures in space of two separate musical tracks could easily be obtained.

PSYCHO-ACOUSTICS

In this system, each sound track is considered as a sound point-source moving in space. In the case of simple geometrical figures (circles, squares, etc.), tests have shown that practical motion frequency cannot exceed 2 or 3 Hz, for accurate perception, although esthetically interesting effects can be obtained at higher frequencies.

Of course, if two or more figures are created simultaneously, pattern perception will be affected to quite an extent and other phenomenons have to be considered:

A■



B■


C■
The music itself: If each musical track is contrastingly different from the other, pattern perception will be increased. Contrasts can of course be of timbral, rhythmic or melodic nature.
Spatial isolation: Giving each track its distinct area in the cube, (ex.: one track rotating at the ceiling and another on the floor).
Pattern contrast: Using contrasting shapes of balance figures (ex.: circular motions opposed to linear designs...).

Location recognition can also be helped with a special «center proximity» control voltage available from the gain controller module. It can be used to control a wide variety of musical and acoustical
parameters such as reverberation and phase, which are closely related to moving sound sources.

Tests have shown that panning prerecorded music will not prove as effective as if the music and sound location have interdependent composition characteristics. In most experiments conducted up to
date, both musical tracks and balance control voltages have been generated simultaneously.

 

  

No funds, no conference

I received this letter from the government explaining why we would not get the funding for the trip to the conference.
They explain that my demand falls between two jury sittings, but they point out that they won't fund english conferences held in a french country.

 

Gouvernement du Québec
Ministère des Affaires
Intergouvernementales

Québec, le 8 avril 1982

Monsieur Daniel Laberge
83xx des Belges
Montréal, Qué.
H2P 2A9

Monsieur,

J'accuse réception de votre lettre du 22 mars dernier relative à une demande de séjour à Paris, dans le dessein de participer à un colloque international. Voici quelques considérations qui m'ont orientée vers la décision exprimée plus loin.

Tout d'abord, le budget affecté aux colloques internationaux s'avère plutôt mince ces dernières années. Aussi, dans un souci de justice, le ministère fait appel à des jurys qui siègent de temps à autre pendant l'année académique. Le dernier s'est tenu le 4 mars dernier et le prochain aura lieu dans la première quinzaine de juin. Votre demande arrive malheureusement trop tard pour obtenir cet avis des experts.

Par ailleurs, je ne vous cacherai pas que le ministère se montre très réticent, surtout depuis le dernier colloque international tenu à Montréal sur l'emploi de la langue française dans le monde scientifique, à subventionner la participation à des colloques tenus en langue anglaise dans un pays francophone.

Pour ces motifs, j'ai le regret de ne pouvoir donner suite à votre requête qui, par ailleurs, présentait sans doute des aspects scientifiques valables.

Dans l'espoir que vous pourrez quand même participer à ce colloque, je vous prie de croire en mes sentiments les meilleurs.

 
 

LOUISE BEAUDOIN
Directrice des Affaires françaises

 

  

Octophony

The soundstage of choice for the electronic composer

3D sound now

After over 30 years, octophony has not progressed the way I initially expected it would.
Many people say they are doing «octophony», but their eight speakers are at floor level, usually forming a circle.
Though there is a core of real tridimensional octophony enthusiasts, their works remain unknown.
Some software plug-ins such as «SpatCube» and «Anima cube» seem promising, but few programs can incorporate their capabilities.
Nowadays, the concept of a 3D soundstage has been replaced with two-dimensional acoustic stage (or scene) replications such as those found in movie and home theatres.
While joysticks abound, there still are no tridimensional transducers on the market.

Acoustic and electronic music

I am a classically trained composer.
The type of music I write, perform and enjoy the most is executed on a stage where the musicians stay in place.
Octophony is not well suited to reproduce these performances because it does not try to replicate an acoustic soundstage.
In octophony, each voice (or instrument) has its position in 3D space and it can move around freely.
Its motion, and the choreographies executed by several voices, become part of the presentation.
Electronic sounds are more appropriate for this type of composition because their parameters can be controled and linked to positional values.
Furthermore, music that is composed specifically for octophony, with spatial movement in mind, will undoubtedly prove to be more effective.

Surround sound

Surround sound is conceived to reproduce live acoustic performances and the ambience of any location.
All the loudspeakers are at the same level, except for one (used for low frequencies only), so it is only two-dimensional.
Even if, in better systems, four channels are dedicated to quadraphonic sound, the loudspeakers are seldom positioned to form a square.
I personally consider surround sound as a worthless gadget, devised to thrill the audience.

My loudspeaker problem

In hindsight, I realize I was foolish to think that the university would supply eight loudspeakers in the manner this project needed them.
After the four loudspeakers were stolen during our tests, I decided to put together my own system. I bought and assembled eight loudspeakers, but I also needed eight amps. Printing the circuit boards for them was as far as I got.
In those days, headphones were not what they are now.
Everyone is wearing them nowadays.
People have gotten used to listening to their music in solo.
In the case of octophony, had I thought at the beginning of the project to build an eight speaker headset, I could have tested it much more.

Octophonic heafset

Octophonic headset

Thanks to Hubert Caron

I only have good souvenirs of working with Hubert.
His dedication, professionalism and intelligence made this project possible.

The future of octophony

Octophony is not an artificial setup like surround sound.
It is the natural extension of stereophony.
Octophony will inevitably develop with time.
Shortly after the end of our project, the MIDI (musical instrument digital interface) specification was launched and it is now the best way to link computers to musical events.
MIDI can already control the x axis for each instrument with Controller #10, but has no provision for the y and z axes even in the Level 2 specification.
With two new controler numbers, more people would be inclined to incorporate the new dimensions.
A 3D transducer is still needed.

The best soundstage
for electronic music

As an acoustic composer, I wouldn't touch octophony.
But, when I wear my electronic composer hat, I find the stereo soundstage incredibly limiting.
There is no doubt in my mind that many other composers feel the same frustration.
Octophony is still in its infancy.

 

  

daniellaberge.net                   

 

 
 
Music section menu
Rhythm explained - Including 27 rhythm exercises The web piano - Play it with your mouse Classical music accompaniments - For winds and string instruments Octophony - The first octophonic system IKO - The electronic trio The Perfect Now - Happy music MollyBelle and Dax, the dynamic duo Flute on a street corner - Music with scores Daniel Laberge's website