午夜福利1000集合

JOE POMPEI bends down and picks up his favourite toy. It鈥檚 a thin black disc
about half a metre across which he holds in two hands like a barber鈥檚 mirror. He
flicks a switch on an ordinary-looking amplifier, and you become vaguely aware
of a tinny beat, as though there鈥檚 someone on the other side of the room
listening to salsa music on leaky headphones. Then he swings the disc towards
you, until you鈥檙e facing it full on, and something extraordinary happens. The
air around your head explodes with music. It鈥檚 as if the Buena Vista Social Club
has sneaked up behind you and burst into action. Pompei motions for you to walk
towards the disc. You take a few steps forward, and the music follows you like a
swarm of bees. Then he tilts the disc away and the sound blips out of
existence.

Pompei鈥檚 toy is a new audio technology that can fire thin beams of sound with
the precision of a spotlight. Focus it at someone鈥檚 head and you can beam them
their own personal music or talk to them in private from the other side of the
room.

Move the beam a few centimetres to either side and the sound shimmers out of
existence. Pompei, a researcher at the MIT Media Lab in Cambridge,
Massachusetts, likes to demonstrate his device by firing sound effects at
unsuspecting members of the public. Waitresses and shattering glass are his
favourite combination.

But the sound beam is much more than just a plaything. It
promises to be the first significant development in audio technology since the
moving coil loudspeaker was invented in 1925. And there鈥檚 a huge market for the
product. According to Woody Norris of American Technology Corporation, a San
Diego-based company which is designing a similar device, sales of conventional
loudspeakers bring in more than $11 billion a year in the US alone. Yet
sound beams do everything speakers can do, and more. Every seat on an airliner,
for example, could have its own beam, eliminating the need for headphones.
Drivers could tune in to the travel news while their passenger listens to
classical and the kids rock out in the back. Movie soundtracks would be
broadcast in glorious technisound, and public address systems would be able to
fire individualised messages. The armed forces could relay signals without
breaking radio silence, or beam sound to distant locations to confuse the enemy.
And the security services could pick you out of a crowd to warn you that you鈥檙e
being watched . . .

Under normal circumstances, sound waves don鈥檛 travel in
narrow beams. Instead, they spread out in all directions from their point of
origin. That鈥檚 why the air around us is full of sound, and why you can hear the
music on your hi-fi without having to stand directly in front of the speakers.
Sure, it鈥檚 possible to direct sound a little鈥攚hen you cup your hands
around your mouth to shout to someone a long way away, for example. But how do
you go about directing a sound with the precision of a spotlight?

The answer is that sound waves are, in fact, directional鈥攊t鈥檚 just that
you rarely notice it. To a reasonable approximation, the 鈥渂eam angle鈥� of a sound
is determined by its wavelength. The smaller the wavelength, the less the sound
spreads out. The problem is, in the audio spectrum, where wavelengths stretch
from about 2 centimetres to nearly 17 metres, all but the smallest waves spread
out 360 degrees.

Wavelength, though, isn鈥檛 the only factor that controls the spread of sound.
Beam width is also determined by the aperture of the source鈥攁 bigger
loudspeaker focuses sound in a smaller area. Make the speaker several times
bigger than the wavelength of the sound you want to transmit and you get a very
focused beam. Of course, that鈥檚 not much help if you want to make a useful sound
beam. Even the shortest audible wavelengths would need a speaker of around 10
metres, and to squeeze the longest wavelengths into a beam, you would need a
loudspeaker the size of a skyscraper.

But what if you converted your sound to really tiny wavelengths, ones outside
the audible range? With wavelengths in the millimetre range, it鈥檚 easy to
generate thin beams using speakers less than half a metre across. And that鈥檚
exactly what audio spotlights do. Pompei鈥檚 disc, for example, is an ultrasound
speaker that fires carefully selected wavelengths into the air.

Hang on, though. There鈥檚 not much point in broadcasting beams of ultrasound
when nobody can hear it. That鈥檚 where an unexpected twist comes in. It turns out
that air spontaneously converts the ultrasound into audible sound. It鈥檚 a
strange nonlinear effect, which mathematicians are only just getting to grips
with. But it鈥檚 what makes sound beams possible. The overall effect is a long,
narrow column of ultrasound extending out from the front of the disc like the
beam from a searchlight. The column acts like an airborne loudspeaker,
generating audible sound along its length. And that, Pompei, says, is the secret
of the sound beam. A long loudspeaker has the same effect on sound as a wide
aperture: it keeps it on the straight and narrow, so that almost all the audible
sound is confined within the beam. 鈥淭he edge doesn鈥檛 go straight from full sound
to zero, but it鈥檚 only a matter of centimetres,鈥� says Pompei.

The origins of the device date back to the 1960s, when sonar researchers were
firing pulses of ultrasound into water and noticing that the pulses
spontaneously converted into narrow beams of low-frequency sound. Realising
they could exploit this phenomenon to generate directional sonar, engineers
scrambled to solve the maths governing the phenomenon. This culminated in 1965,
when Orhan Berktay, then in Britain at the University of Birmingham, published
an analysis now known as Berktay鈥檚 equation. What he found was that water
distorts ultrasound signals in a complex but mathematically predictable way.

Music in the air

Berktay was interested in underwater sound propagation, but his work stirred
up a wider debate within acoustics. Could the same effect happen in air? Would
it be possible to reproduce complex sounds such as voices and music in a sound
beam? According to David Blackstock, an expert in non-linear acoustics at the
University of Texas in Austin, there was widespread disbelief that the trick
would work in air. 鈥淥rdinary experience tells us that sound behaves very
linearly,鈥� Blackstock says. 鈥淚f you go to a concert, you hear just as well far
away as you do close up. But that鈥檚 because the non-linearity is weak. You don鈥檛
usually notice it.鈥�

In 1975, Blackstock and a graduate student called Mary Beth Bennett decided
to test the idea. They modified an underwater transmitter and fired inaudible
tones into the air. Just as they anticipated, the pure ultrasound tones
interfered with one another to create an audible tone, proving for the first
time that air, like water, propagates sound in a non-linear way.

The experiment suggested it would be possible to produce a sound beam by
firing pulses of ultrasound and letting the air do the rest. But it didn鈥檛 point
the way to a useful audio spotlight. Complex sounds such as speech and music are
not the same as the pure tones used by Bennett and Blackstock. They contain
multiple frequencies and wavelengths, all of which interfere with one another to
produce a smorgasbord of distortion. Selecting the right ultrasound frequencies
to generate coherent audio is tremendously complicated.

Enter Masahide Yoneyama of the Japanese electronics company Ricoh. In the
early 1980s, he realised there was a straightforward way to convert a complex
mix of sound into an ultrasound wave. His insight didn鈥檛 go all the way to
sorting out the distortion problem, but it did identify the technique that makes
audio spotlights possible. It鈥檚 called amplitude modulation, and it has been
used for decades to broadcast medium wave radio.

The basic trick of amplitude modulation is to take a pure-frequency wave,
called a carrier, and multiply it by the signal you want to transmit. This
creates a hybrid wave鈥攖he carrier with the signal superimposed on it
(see Diagram).
In the case of radio this is an electromagnetic wave that propagates
long distances through space. When this wave reaches its destination鈥攁
correctly tuned radio set鈥攊t is demodulated to regenerate the signal.
Exactly the same can be done with sound. Mix speech or music with a pure-tone
ultrasound carrier, and the result is a hybrid wave consisting of the carrier
with the audio signal superimposed.

So far, so good. This trick gives you an ultrasound wave that you can
broadcast as a very narrow beam. But what happens when it moves through the air?
It turns out that amplitude modulation creates two new sets of frequencies, one
slightly higher than the hybrid wave and one slightly lower. And if air works
like water, then according to Berktay鈥檚 equation these sidebands would interact
with the hybrid wave, much as Bennett and Blackstock鈥檚 pure tones interfered
with one another. One result of this interference would be a sound wave
identical to the original, unmodulated signal. It should be possible, Yoneyama
reasoned, to use this approach to create a sound beam.

So Yoneyama and his colleagues at Ricoh joined forces with Nippon Columbia to
build the device. In 1983, they set up a large ultrasound speaker and
transmitted amplitude-modulated ultrasound through it. Just as they expected,
they heard their original sound reproduced inside a beam that extended out from
the face of the speaker. But they also got a huge amount of distortion. And the
distortion became worse as they turned up the volume. The overall effect was
rather like listening to a full symphony orchestra through the speaker of a
tinny transistor radio.

What went wrong? Yoneyama was correct that the sidebands would interact with
the hybrid wave to reproduce the audio signal. But he missed a crucial fact.
Berktay鈥檚 equation actually predicts that the sound produced by this process
will have two components. One of them is the original sound wave, but the other
is a dreadfully distorted wave. What鈥檚 worse, though the volume of the original
sound wave rises in direct proportion to the volume of the ultrasound, the level
of the distorted component rises exponentially. At volumes louder than a whisper
the distorted component simply swamps the original signal. Faced with an
impossible choice between distortion and inaudibility, Yoneyama abandoned the
technology.

But according to Pompei, the Japanese researchers missed a trick. Instead of
giving in to the distortion, he says, they should have turned it to their
advantage. Forget the standard signal that gets swamped out of existence. What
you really need to work with is the other, distorted one. The powerful one that
rises exponentially when you crank up the ultrasound volume.

How to make use of this distorted component? Pompei realised that, thanks to
Berktay, the way the sidebands interact is entirely mathematically predictable.
So all you have to do is start with your sound wave, modulate its amplitude to
turn it into ultrasound, and鈥攈ere鈥檚 the clever bit鈥攔everse distort
it. In other words, you use the Berktay equation to work out what air does to
ultrasound to create the unruly distorted part of the signal, and then do the
exact opposite to your signal. It鈥檚 like stretching a short, fat man until he鈥檚
tall and thin, then taking him to a hall of mirrors. The concave glass that once
squashed him out of all recognition now restores him to his original shape.

Of course, when you pass the result back through the air, you still get two
components. But now, the one that gets swamped out of existence is the one
that鈥檚 no use to you. And the one that produces your original sound is the
previously unruly exponential component.

Using this method Pompei can produce almost distortion-free sound at 80
decibels鈥攁bout as loud as a vacuum cleaner. 鈥淚t鈥檚 not perfect, but it
reduces distortion from 50 per cent to around 5 per cent,鈥� he says. That鈥檚 not
far from ideal. The best conventional loudspeakers generate less than 1 per cent
distortion, and Pompei says you need 鈥済olden ears鈥� to detect anything less. 鈥淚鈥檓
getting closer to 1 per cent,鈥� he says, 鈥渂ut I need more maths.鈥�

The technique does have a few drawbacks. The biggest is that reversing a
sound wave through Berktay鈥檚 equation generates an infinite number of harmonics.
In theory, all of these must be transmitted to generate clean sound, although in
practice the outer ones are so weak that five or six will do. Even so, the
signal includes a vast range of frequencies, requiring a very high-quality
ultrasound speaker. These just didn鈥檛 exist until a couple of years ago, and
Pompei says building one was just as challenging as cracking the maths. But with
several patents pending, he won鈥檛 give any more details.

The harmonics problem is so great that American Technology Corporation has
abandoned the Berktay equation altogether. Joe Norris, the company鈥檚 resident
physicist, says it has developed an alternative mathematical technique. 鈥淭here鈥檚
more than one way to get the right waveform,鈥� he says. But he won鈥檛 say how it鈥檚
done.

Another problem is that the process reduces the volume of the bass
frequencies. 鈥淭he low end is the hardest,鈥� says Norris. 鈥淵ou get a 12 decibels
roll-off [decline in volume] per octave.鈥� What that means in practice is that
frequencies below 200 hertz are all but impossible to reproduce inside the sound
beam.

Pick your spot

But that鈥檚 a small price to pay for directing sound with the precision of a
flashlight. At 10 metres, Pompei鈥檚 device can pick out a spot about half a metre
in diameter. Even at 50 metres the sound is confined to a circle 2.5 metres
across. That鈥檚 great if you want to communicate privately in a public space, or
send a message across a room without disturbing other people. It鈥檚 easy to think
of uses for the technology, some of them sinister. Pompei laments the fact that
his name crops up on numerous conspiracy theory websites accusing him of
developing mind control technology. He prefers to think of football coaches
relaying messages to their players.

And there鈥檚 more to audio spotlights than directivity. The devices developed
by Pompei and American Technology Corporation promise to completely change the
way we record and reproduce sound. By choreographing moveable beams, we will be
able to recreate the real world of sound in our homes and cinemas.

In real life, people鈥檚 perception of sound is highly directional. Stand in a
busy street with your eyes closed and you can pinpoint the origin of every sound
with unerring accuracy. But if you videotaped the scene and tried to do the same
at home, you鈥檇 always end up pointing at the television. Audio spotlights can
change all that. When an audio beam hits a smooth surface it reflects off it,
like a flashlight beam bouncing off a mirror. But if you make the surface
slightly rough, the beam will scatter. Suddenly, instead of directed sound you
have what seems like a normal sound source, with the waves spreading in all
directions. And it sounds for all the world as if the noise is coming from the
point where the beam hits the surface.

American Technology Corporation is already working on a television that
reflects sound beams off the wall behind the viewer. And Pompei talks about
bouncing beams off a cinema screen so that voices, car engines, gunfire and
other sounds emanate from the spot where the action happens.

Virtual reality audio, however, is probably a few years off. The early
applications will be based on the sound beams鈥� directivity. American Technology
Corporation has signed a deal with shipbuilder Bath Iron Works to install sound
beams on the deck of its new US Navy destroyer. These will let radio operators
listen to their own transmissions while keeping track of the goings-on around
them鈥攁 trick they usually accomplish by putting their headphones over just
one ear. Carmaker DaimlerChrysler, meanwhile, has struck a deal to put Pompei鈥檚
audio spotlights in a test vehicle called the MAXXcab concept truck, allowing
each passenger to listen to their own music without disturbing the others. If
either of those are a success, more deals are sure to follow. You should believe
your ears.