Recently however, it has become possible to mould ultrasonic sound fields to form keyboards made of sound (see, for example, http://www.newscientist.com/article/dn26463-ultrasound-makes-hologram-keyboard-touchyfeely.html). At least, that's what they feel like: the examples currently available use infrared detectors link to pattern recognisers to determine the positions of a user's fingers. Ultrasound beams are then focussed at the fingertips, generating a pressure on them. So in fact the keyboard only exists at the points where it is felt; a kind of relative acoustic object.
In principle, one could generate all sorts of shapes in this way. Different textures could also be produced : the details of such an acoustic sculpture are limited by the wavelength of the sounds used to mould it.
At the cost of many more ultrasonic projecting elements and a lot more processing power, acoustic objects which are there all the time, whether they are touched or not, could be made too. If the idea of invisible but feelable objects seems a bit unsatisfactory, suspended particles are very easy to levitate, so one could make objects from smoke or mist or even glitter.
It's easy to imagine all sorts of applications for such things : art objects, theatre effects, virtual reality games and systems without the need to dress up like a robot, training facilities, control systems.
There is, unfortunately, a "but"... ultrasound is absorbed very rapidly by air, so the objects could be formed only within a very few centimetres of the array of projectors. There are two possible ways round this, however. Ultrasound travels much further underwater than in air, so much larger submerged acoustic shapes could be formed : acting as virtual walls to separate sharks from divers for example, or for fishing. Conceivably, one could also make subterranean structures (ultrasound travels easily underground too), like temporary walls or baffles to support the ground after earthquakes perhaps, or even for tunnelling
Another possibility would be to overcome the main limitations of audible sound by using what is known as a parametric array. This enables sounds of even very low frequency to be formed into precise beams. These beams could be made quite powerful, since they would be silent unless aimed directly at the ear or bounced from a surface (I've added a short explanation of how these arrays work below).
A key characteristic of acoustic pressure-beams, of whatever frequency, is that they can only apply a force in the direction in which they are sent. So, while it is simple enough to make a flat or gently curving shape by using a panel covered by a layer of small acoustic projectors, to make something like a sphere or a hand-shape, there would have to be projectors in every direction - but one could imagine the walls, floor and ceiling of a gallery or wall being coated with such projectors, along with infrared detectors. A great deal of processing power would be needed to work out where people and other inconvenient objects are in the room, and literally work round them to make the shapes - and the projectors and sensors would need to be small, cheap, efficient and precise, but ever-faster processor speeds and ever-better MEMS (MicroElectroMechanical Systems) will soon be able to deliver both.
BACKGROUND : SOUND PROJECTION BY PARAMETRIC ARRAYS
When a sound wave is produced by a loudspeaker (or by any other source), it will tend to radiate in all directions as long as its wavelength is much longer than the loudspeaker (source) size. Hence, if you speak in the open air, your lower tones can be heard by someone standing behind you, because they are a few hundred Hz, corresponding to wavelengths of a few decimetres.
(Frequencies are related to wavelengths through the well-known equation
v = fλ (v : velocity ; f : frequency; λ : wavelength)).
The highest-pitched components will not be very audible, as they are several kHz in frequency, and so around a centimetre in length. Since they are therefore smaller than the mouth opening, they are quite directional.
Underwater, since sound speeds are much higher than in air (typically about five times faster), the wavelength of a wave of a given frequency is proportionately large. This means that, unless sources are many metres across, all audible-frequency underwater sound waves tend to radiate in all directions.
In many applications, it is desirable to generate directional sound beams, for communication or for pulse-echo detection techniques. Narrow beams reduce off-axis noise levels, travel great distances with little energy loss, produce few confusing side-echoes and are more suitable for transmitting sensitive information.
However, in all media, higher-frequency sound waves are absorbed over shorter distances than lower ones, and for this and other reasons one often wishes to transmit relatively low-frequency sounds in directional beams.
An elegant way to make a low-frequency directional sound source is the parametric array. If two sound sources generate waves which differ just a little in frequency, then waves with that difference frequency will be produced, along with others whose frequency is the sum of those of the sources. The wavelength of the difference wave can be a long as required, but it maintains the directionality of its parent waves. Parametric arrays exploit the fact that sound velocity depends on density. At high sound powers, the pressure in the compressions becomes very large, increasing density significantly and therefore speeding up the sound wave briefly; the reverse happens in the rarefactions. The effect of these velocity changes is to distort the wave from its usual sinusoidal form, as shown in Figure 1.
A non-sinusoidal wave is equivalent to a sum of component sinusoids. In the case of the parametric array, these components include the original waves, together with the sum and difference waves: the difference wave being the one of interest (Figure 2).
Mathematically, the directionality of a sound wave is defined through a directivity function, which describes the intensity (or other measure) of the sound as a function of angle from the source axis. For a parametric array, the directivity function is given at the end of this blog.
The parametric array effect was discovered by chance when, in 1951, acoustician Peter J. Westervelt was working at the Office of Naval Research in London. He noticed that an experimental superheterodyne (frequency-mixing) radio receiver was generating directional audible sound at low frequencies. It was not until 1960 that Westervelt was able properly to explain the effect, which he did at a meeting of the Acoustical Society of America, and in a published paper a few years later.
Today, parametric arrays have numerous underwater applications, and they are also used in museums and galleries for the "acoustic spotlight" system in which recorded descriptions are beamed at appropriate exhibits and can only be heard by people close to the exhibits. A parametric array is also the basis of the LRAD (Long Range Acoustic Device), used to direct painfully load sounds at, for example, Somalian pirates. Underground, sound beams from parametric arrays are used for prospecting.