lets appliances speak softly
Technology Research News
In an age when electronic devices like
computers, cellphones and appliances sport speakers and microphones, it
might make sense to use the audio equipment to allow the devices to communicate
not only with humans, but also with each other.
Just imagine, for instance, how nice it would be if the telephone could
tip off the stereo to turn itself down when the telephone started ringing.
There has been an obvious drawback to granting appliances the ability
to talk among themselves, however: the chatter of appliances keeping each
other updated has at least as much potential to annoy as other people's
cell phone conversations. R2D2's beeps might have been cute, but that
was only because the movie didn't last more than a couple of hours.
Researchers from the University of Florida are giving computers with ordinary
speakers and microphones the ability to communicate using sound, but are
tapping into the quirks of human auditory perception to make that communication
silent to humans.
The researchers realized that the microphones and speakers increasingly
included with computers could be used to form a type of network, said
John Harris, an associate professor of electrical and computer engineering
at the University of Florida. These audio networks have some advantages
over current wireless technologies, which use radio or infrared waves
to allow devices to exchange information, he said.
Radio waves work through walls and can transmit over a relatively long
distance. This requires coordinating the signals so that people near each
other don't use the same frequencies. "You have to be sure that all the
codes intermatch and there are no conflicts [or] you can accidentally
turn off everyone's [stereo] when you pick up the phone," said Harris.
The infrared communications widely used in remote controls don't have
the conflict problems of radio signals, but require a line of sight between
Sound waves have the advantage of going around corners, but not through
walls. "Because the buildings that we live in are built for audio, it's
perfect for talking among appliances," said Harris.
The researchers used a pair of quirks in human hearing to make silent
audio communication possible. "Basically we're trying to bury the sounds
the appliances are making below our sensitivity threshold," Harris said.
As sounds become shorter in duration, we perceive them as quieter than
they really are, he said. "If you make a sound short in length, you also
make it harder to hear," said Harris.
Sounds start fading for us when they are less than 100 milliseconds, or
a tenth of a second, long. A 32 millisecond pulse, for instance, seems
to us to be five decibels quieter than it really is, and a two millisecond
pulse sounds 18 decibels quieter.
For comparison, very soft music is about 30 decibels, background music
60 decibels, conversation 65 decibels, loud classical music 80 decibels,
loud rock music 115 decibels and the threshold of auditory pain 130 decibels,
according to Harris.
The second quirk of human listening is that we have evolved to hear certain
frequencies more clearly than others. The researchers kept their silent
sounds away from these frequencies.
Humans can hear sound waves that range from 20 to 20,000 hertz, or vibrations
a second. The lowest note on an 88-key piano, for instance vibrates at
27.5 hertz, while the highest vibrates at 4156 hertz. We hear best, however,
in the 2,000- to 4,000-hertz portion of our speaking range. This is because
we have evolved more acute hearing in that range to distinguish similar
sounding consonants like p and b.
By exploiting both effects, the researchers were able to produce sounds
ordinary computer microphones could pick up but that were below the threshold
of human hearing, said Harris.
They used a 15,000 hertz signal and 8 millimeter pulses to literally make
the sound disappear. Due to the effects "the final sound is less than
0 decibels, so we can't actually hear it," he said.
The researchers' lab demonstration of the effect allows a person to type
letters on a keyboard on one computer and hear those letters spoken by
another computer across the room that is communicating by sound with the
first computer. The computers communicate by tapping out, in the sound
pulses humans can't hear, the binary ASCII codes of the letters.
When the sound rises above the level of human perception, it sounds like
crickets, said Harris. The communication doesn't require any equipment
beyond ordinary microphones, speakers and desktop computers; the compute
power it needs is within the range of an ordinary cell phone, he added.
In the demonstration, the computers communicate at the relatively slow
speed of 200 bits, or pulses per second. Each letter is represented by
an 8-pulse byte. "It's meant to be very low bandwidth and we haven't pushed
the communication [speed]. You're not going to be transferring files...
with this, [just] information about what the time is and things like that,"
Machines could be talking among themselves within a few years, said Harris.
"As speech interfaces become more common, our nonaudible audio communication
system will be very easy to implement since mircrophones and speakers
will be everywhere," he said.
The researchers are currently working to incorporate the system into actual
appliances like lamps. "We're going to... see if we can create some smarter
types of appliances," said Harris.
Harris's research colleague was University of Florida graduate student
Paul Baker. The research was funded by the University.
Timeline: 2-4 years
TRN Categories: Signal Processing; Human-Computer Interaction
Story Type: News
Related Elements: None
Split pulses speed signals
Gender gap shows
Software lets appliances
gains light throttle
memory does not fade
Research News Roundup
Research Watch blog
View from the High Ground Q&A
How It Works
News | Blog
Buy an ad link