1. Introduction In this article we shall put in writing some of the ideas and considerations that guided us in creating an installation that attempted to use sound in an architectural context. Architecture is an art that is closest in its essence to the problems of space and space perception, with the treatment of space being not limited to geometrical aspects only, but rather has to do with the function of the space, its flow, hierarchy and many other criteria. The ability to perceive space and distinguish between qualities of different locations is achieved therefore by the ear no less then by the eye.

In many cases, we have preconceived notions of space that we acquired by long term living experience in these spaces. For instance, the distinction between public/private function enters our conscience via multiplicity of factors such as the size of the room, the height of its ceilings, the division of interior space, type of building materials, furniture design and many more. All these elements have a crucial effect on the auditory quality of the space, such as long reverberation times and typical reflections of the public spaces. These could be simulated by electronic means such as commercial effect processors.

In addition to these "learned" acoustics-space aspects, there are notions of space which are not so readily related to acoustics but nevertheless can be controlled or at least influenced by sound. In our project we attempted to address several such notions - hierarchy, space flow and connection between remote locations - aspects which seem not to have clear sound associations but are nontheless susceptible to control by the use of sound as an active design element.

In our artwork we attempted to combine architecture and sound by creating a link between places by means of sound rather then by using architectural methods or direction signs. In the past it was obvious that the source of sound comes from a central location - the Christian church bells or the Islam "muezzin" calls from mosque towers - the sound was a unifying factor that brought together all houses in the surroundings into a single village. Today also many homes have bells that hang at their main door and inform the inhabitants about the happenings at the entrance. In this work we have applied the same idea: by communicating a sonic presense originating from one focal location to various distributed sites in an architectural complex, a unification of spaces is achieved.

2. Description of the installation One variant of the installation was realized recently at the Hong-Kong University of Science during the International Computer Music Conference held there on August 1996. At the entrance plaza of the university an existing Sun Dial statue is located. By attaching chords from its top to basis we have converted the Sun Dail into an ancient music instrument - the wind harp, and by means of modern elements such as sensors and computer, the information about the trembling of the chords was converted into sound and transmitted to the interior. In the interior hall, the sound was played sequentially through an array of 16 speakers, which produced an effect of sound motion. The speakers were aligned along a main wall of the hall, with the relative motion of the sound designed so that a sense of direction was created from the plaza towards the interior.

The above setup achieves several architectural goals: When standing in the interior space of the university one is aware of the happenings at the entrance plaza by hearing the sounds that were triggered by the wind outside. This interactive sound information creates a link between two separated locations. The sound motion gives the sound a role of a compass that constantly transmits the sense of direction relative to the central plaza. Moreover, the sound motion increases the space flow by creating virtual fast moving objects that "fly" past the listener.

3. Sound design considerations and realisation. As was mentioned above, the wind was used as a trigger and not as an actual sound source. The sounds were generated by reading sound files from a database that was held on the computer disk. We have used 24 sound banks with 3-6 sounds in each bank (over 100 sounds in ~250 MB total disk space). The sound files were recordings of sounds from natural and synthetic sources, which could be roughly described as sound textures - a relatively long (1.5 - 5 seconds) acoustic events with inner complexity of mostly static nature, but some having also a certain sense of development. We have found that long sounds were needed to convey the desired sound motion, and the internal variability of the sound helped to create a coherent feeling of a moving sound.

In our setup we have use six sensors with each sensor assigned either a distinct sound or pitch shifted versions of a single sound assigned to two sensors at most. The timbre of the sound changed according to the time of the day (by loading a new sound bank), thus preserving the original Sun Dial's meaning. Our experience showed that the density of the sound was better controlled by adding different sounds rather then by repeated triggering of the same sound. Since the strength of the wind corresponded to triggering of more sensors, we achieved by the above method a better correspondence between the sound density and the wind activity.

4. Some problems and future improvements. In this section we would like in short to present some of the problems that we encountered and some ideas for improvements.

a) In order to convey a better sense of space continuity, the sound presense should be much more significant. This can be achieved by having more speakers and extensive diffusion of the sound around more locations. It is also psychologically important to put the sound into a significant role in the complex rather then risk to treat it as a local attraction.

b) The control of the sound motion speed seems to be important. In our setup we have calibrated the sound motion once, so as to obtain an optimal sense of motion. We believe that a grater variety of effects will be achieved by controlling the speed parameter also.

c) Apparently some sound textures sounded better then other in this context. Sound of bells and other metallic sounds gave probably the best feeling of motion and had a good presense in the hall. Trembling strings and other more gentle textures were not as effective, with a worse sound motion percept. It was surprising that complex textures that normally have a strong presense, such as loud jet sounds or rain/water sound and other similar sounds, did not create a good result. Their sounds seemed to fill the space with some kind of "static noise", which had little presense due to blending with the natural ambience, and having a poor sound motion. It seems that we still need to have more understanding of these phenomena.

5. Conclusion David's Harp ("Kinor David" in Hebrew) is an interactive installation which attempts to unify separated spaces in architectural complex by means of sound. Communicating the sonic presense that originates at on central location due to some local activity such as blowing winds, an awareness to the outside environment is created. This bring the outdoor life into the interiors of the building and bind the two together. In some sense it is an abstractization of the "bells at the entrance of the house" idea, only that here it has received a proper scaling so as to fir the sizes of a large architectural site and also reflect the technology of our times. In addition to interactivity, another important element was the sound motion. The role of sound motion is twofold: functional - as an active compass that constantly transmits the direction to the central location, and as an active design element, lively and creating the required space flow within a building.