Page  183 ï~~Using Hyper-instruments for the re-distribution of the performance control interface Tim Anderson Debbie Hearn Dept of Electronics, University of York, York Drake Music Project Research Group 3 Ure Lodge, Ure Bank Terrace, Ripon, N. Yorks Abstract 'Hyper-instruments' are large scale musical systems which enable the control of complex musical events by a performer. They can thus allow various aspects of the performance interface and its relationship with the operation of the musical system to be altered. This can allow a redistribution of the performance skills required which is especially relevant to disabled musicians - augmenting their motor skills when performing musical material within a pre-constructed framework. Hyper-instruments thus facilitate a moveable boundary between preparation and performance according to the ability and experience of each person. 1. The use of Hyper-instruments to re-distribute the performance interface All musical instruments require some kind of physical interaction between the player and a performance interface into the instrument. This interaction is now practised with acoustic instruments as a highly developed conventional art - a complex relationship which transforms physical actions into musical sound. 'Hyperinstruments' consist of large scale musical systems which can be designed to respond to performance control in a complex manner. A complex sonic response may thus be generated by a musical performance which can involve unconventional combinations of gestures and modes of control. Such a system is acting as a new, more complex instrument. This makes them highly suitable for use by disabled musicians who typically have different combinations of musically precise movements to a non-disabled performer. Such systems can facilitate the control of musical events which are more complex than could be managed by human performance actions, and can thus allow various aspects of the performance interface and its relationship with the operation of the musical system to be altered. Such modification of the performance interface facilitates a re-distribution of the performance skills required. This redistribution is especially relevant to disabled musicians - augmenting their motor skills to facilitate the performance and control of musical material within a pre-constructed musical framework (whether performing live or as part of a composition process). The concepts involved in the data processing modules used in constructing such systems are well known: this paper describes a use of such flexible high-level performing tools to enable the performance interface to be customised. For example, the number of degrees of freedom required for a particular physical performance gesture can be deliberately reduced, but this reduction can be compensated for by an increase in the precision of the movement. This increased resolution can then be used to control pre-prepared material or processes within the instrument which can increase the density or complexity of musical output for a given control input, thereby replacing the musical interest which might be lost by a reduction in other aspects of performance control. 2. Construction of Hyper-instruments 2.1 Hyper-instrument structure Hyper-Instruments consist of three notional components: (a) Some kind of physical interface for the input of control data from the performer into the instrument; (b) Processing of this performance data; (c) The generation of sound output by musical processes and structures, or sound processing - controlled by the performance data. 2.2 MIDI and modular systems The MIDI communication standard has enabled the codifying of elements of a performer's actions in a standardised manner. MIDI is an appropriate communication medium for a single performer, with (just) adequate performance bandwidth, and the advantages of standardisation. Thus, proprietary subsystem components (inter-communicating via MIDI) can be mixed and matched as appropriate to construct a Hyper-instrument. Components will include: (a) software to store, generate and process data; (b) synthesis and sound-processing devices; (c) MIDI performance controllers; (d) computer data entry devices which are not often considered as performance tools. This paper thus also acts as a review of practice which is developing the expressive capabilities of MIDI in performance. 2.3 The processing and music generation subsystems A Hyper-Instrument's musical processing control aspects may exist as a single (non-distributed) system: either custom-built hardware, or a custom-designed complex software application, whether created using higher-level environments (eg Max, Sybil, Interactor), or a more general purpose language with musical extensions (eg MidiLisp, HyperMidi). Alternatively, a Hyper-Instrument can be constructed by assembling a number of disparate simpler components which are readily available. These may be hardware processors (eg MEP4), or facilities available within proprietary software applications (eg MidiGrid, Cubase). Each has its advantages: ICMC Proceedings 1994 183 Interactive Performance

Page  184 ï~~The latter allows relatively easy adaptation (in the field) to each performer's requirements and abilities, and enables modification as a performer's abilities (physical or 'musical') change, eg with practice. This facilitation of adaptation is important if the aim is to provide Instruments which can be used in the world of music at large (rather than inside institutions with software and technical support) by disabled musicians. Such systems need to be maintainable, cost-effective, and - most importantly - be understandable and alterable by musicians themselves, or their assistants. In contrast, a typical unified software solution will usually be less amenable to performers do not have considerable programming ability - a relatively rare combination. 3. The components of a Hyper-instrument 3.1 Performance data interface Input devices can include: (a) Computer input devices such as the tracker ball, mouse, joystick, drawing tablet, touch-screen, or the console keyboard itself. These can be used with the foot, or with a very small but precise movement of the hand. Infra-red head direction detectors to emulate mouse movements are also available; (b) Various types of switches, pressure pads of varying sizes (mm to m) and sensitivities (cm to pm); (c) Pressure transducers (eg piezo-electric "bugs"); (d) Proximity detectors or light intensity detectors; (e) MIDI performance controllers, typically using single notes or regions on a MIDI keyboard, or pads on a MIDI percussion controller. Again the foot is often used. (f) Ultrasonic distance sensors (eg 'Soundbeam' 'Midi Gesture') which produce MIDI events from movement within or into the beam. The range of discriminable movement can be m to mm. 3.2 Processing of input performance data The input data (eg from various physical movements or gestures) can be processed in a variety of simple ways such as (a) filtering. For example, a single direction of movement may be 'parsed out' from a two-dimensional movement of a tracker-ball (ie movement on one of the axes will be ignored); (b) scaling, or other simple arithmetic operations; (c) merging of data from different performance inputs (from the same or different performers) into a single piece of musical control data. For example, one performer can trigger notes at the right time, while another controls the pitch (say one of 8) by position in a sound beam several metres long. A third performer could be moving a finger within a 2cm range over a proximity detector to control the velocity, timbre or volume. More complex (typically software-based) analysis of speed direction or acceleration can also be performed, as exemplified by the Radio Drum. For example, an event can be triggered by a reversal of movement. 3.3 Control of music and sound output Various operations can be carried out to convert performer actions into musical output: (a) Single events (notes, chords, phrases etc) can be pre-composed and stored. These can then be selected and triggered by performance actions (as described above). However, there may be an element of predictability in being restricted to triggering fixed material; (b) Pre-stored material may be mutated. Eg, notes can be transposed, chords can be transposed or inverted, and phrases can be transposed, inverted or retrograded, or different sections triggered, or looped etc; (c) Additional material can be added, for example creating additional notes with various fixed or variable relationships with the 'input' note. Such alteration may be pre-selected or controlled by performance inputs, whether from another performer or another aspect of the performers movements; (d) Timing can also be synchronised to other material being played (perhaps by others in the group, or other phrases etc being played at the time within the same Instrument). For example the onset times of triggered material can be quantised to fit with other events being played (eg by MidiGrid); (e) Audio processing of the sound can be performed. For example the selection of reverberation style, chorus, or other 'effects' can be controlled, or the parameters of such effects can be altered. 4. Conclusions - the music preparation / performance boundary Hyper-instruments allow the boundary between preparation work (pre-composed phrases, chords and structures, and a plan of their performable interrelations etc) and performance work to be moveable according to the balance of performance and compositional ability and experience of a person. A more severely disabled performer can spend more effort in an 'off-line' preparation of material and systems to enable the performance to then proceed with fewer or less precise physical performance inputs. However, the more complex the instrument, the more experience and awareness is required by the performer of the effects of each input action. In every case, musical choices are required, and care and skill in instigating them is necessary in order to create a musically coherent and satisfying result. For example, even if the onset times of triggered events is heavily quantised, the choice and rough time of events is still significant. Thus, software and hardware elements of Hyper-instrument systems allow the musical effect of performance gestures to be more complex. However, the feedback of the musical output in influencing the performer's appreciation of the relationship of the musical output with his/her performance input is still vital - such systems can still produce boring results if they receive unconsidered performance input. For example, as described above, several performers can interact to jointly produce a performance - each has fewer physical parameters to consider, but musical awareness and ability is still needed for a 'good' performance. Such instruments may still be as difficult to learn to play well - the overall performance bandwidth of the interface is maintained - but its components are re-allocated to different aspects of performance control. Thus, complex music can be performed by musicians with a wide variety of types of abilities. NB. Examples, diagrams and references have been omitted from this paper by force of space restrictions. A full version is available. Interactive Performance 184 ICMC Proceedings 1994