Universal Translators - undoing babel
- From: Willie.Mookie@xxxxxxxxx
- Date: Fri, 08 Jun 2007 02:02:10 -0000
I've been a fan of Star Trek since it first came out in the 60s. I
have not followed it closely after TOS, but have now grown fond of
ENTERPRISE with Captain Archer and crew. Its funny to see
developments over the years, from cell phones to powered doorways and
such - and as an engineer I find myself sometimes thinking how this or
that detail might work
One thing that seems to be just a conveneince for advancing the plot
is the universal translator. haha.. That would NEVER be built,
there's no way it could work... that's the hairy knuckled engineer in
me who has to actually build this shit - haha.. but, as I was flying
from a meeting at White Sands the other day - thinking about 3-D
display systems - and observing the exit signs before take off
(printed in several languages) it occurred to me that a 3-D
holographic display - could present a different sign to every person
in the plane - because observing the sign from a specific seat would
observe it from a specific angle. and that angle could be encoded in
the hologram to encode it as a specific image. That is, each seat
could see the same sign, from a different angle, and the hologram
would display a different sign.
Now in holography the different angles are all related to create the
illusion of a 3D object. However, there is no reason a computer
generated hologram should be like that. An active holographic display
screen could display chinese for one seat, korean for another, english
for another, french for another and so on.
Now, take it a step further. Imagine that each person has on their
person some sort of RFID chip that identifies them, and their primary
language. So, when they sit at their seat, the seat knows the
language they speak and read and write. So, that information could be
sent to ALL the signs in the plane, to present at that angle, for that
seat, signage that that particular person can read. People in other
seats would be treated similarly, so the same sign from different
seats would be 'translted' for each of the occupants.
NOw, take it a step further. Imagine that the plane is equipped with
a set of sensors that track the movement of an individual and present
appropriately labeled signage to that person no matter where they are
located in the plane. So, from the moment they walk in to the moment
they leave, they see signage that is appropriate to their primary
language... So, someone looking at a display on a monitor for
example, would see English if that was their primary language - but
someone else standing along side of them, seeing the display at a
slightly different angle, would see it as French or Korean say.
Tihs is doable with 3-D technology and RFID technology today. It
might be common place in 3 to 5 years if there was a demand for it.
Now, this is with visual holograms. What about sound? Well,
holograms are created by the interference of waves in a photographic
medium. The interference pattern diffracts beams of light creating
the wave front of an entire or whole image - holo-gram. A computer
generated hologram computes the interference pattern and uses adaptive
optics to create a dynamically changing hologram to create any sort of
3-D image. And that's the basis of the displays I just spoke of.
Well, sound waves can be interfered and diffracted as well.
A discrete version of this is phased array technology. Each point on
a surface can be made to emit waves in all directions - but by
controlling the phase of each of the points, the waves can be made to
interfere and add and cancel in ways that recreate a very complex
pattern - very similar to a hologram in many respects - except oow the
computer is computing the phase and intensity of each point in an
'image' at each point in time. This is typically too slow to do with
light -using current technology- but it is routinely done with
microwaves. And there's no reason it couldn't be done with sound
waves.
And just as phased array pattern of waves can be generated - they can
be picked up - and analyzed - and complex 3-D patterns of sound can be
recorded and reproduced with a phase array system.
So, this is an analogue of the holographic signs.
the trouble with this is that you need very large arrays of
microphones and speakers in order to get any resolution. No worries,
there's a trick that's possible.
FM radio beats two frequencies against one another to get a resultant
frequencie without noise. You can do the same thing with sound
waves. You can take two very tiny ultrasonic beams and cross them in
the air, and they will beat against one another producing a
resultant. Fetal heart rate monitors work like this. They bounce a
ultrasound wave off a beating heart, the doppler shift in the return
pulse interferes with the outgoing pulse and produces an audible
resultant.
So, two supersonic beams with slightly different frequencies can cross
at a point in space, and produce an audible sound from that location -
whose phase and intensity is accurately controlled.
In short, a tiny device with two phased arrays of ultrasonic emitters
can fill the room with very complex audible patterns. The ultimat
stereo system.
But by knowing where everyone's ears are - aking to knowing where
everyone's eyes are - a computer can produce sounds for each person
consistent with their native tongue.
Unfortunately you can't use this trick to make microphones. No
worries. Oxygen and Nitrogen in the air is magnetic. You can take
liquid oxygen and attract it with a magnet.
http://demoroom.physics.ncsu.edu/html/demos/188.html
This may be useful in detecting air movement or oscillations in the
air with microwaves. That is, instead of having an array of sensitive
microphones covering the ceiling and analyzing the array as a phased
array of sound recievers - again, you may be able to generate an
intersecting pattern of microwaves - that create a pattern of
detection points in air itself by looking at the reaction of a
standing microwave pattern to the changes in impedence physcal motion
caused by sound waves create. This is a very sophisticated doppler
radar.
A 3-D array of virtiual microphones would have other uses as well.
One could monitor the motion and quality of air. One could listen to
the opeation of equipment and interpret it. One could detect leaks,
air pressure and th elike. All useful stuff - with nearly a point
like detector - based on an integrated circuit that emits and detects
two patterns of microwaves and then analyses the impact of air motion
on them - using a computer to create a detailed image of the sound
waves
http://smirc.stanford.edu/
So, now we have the basic input-output mechanism for a universal
translator.
We have a tiny device that can using two separated ultrasonic phased
array emitters (and these could be the front and back surfaces of a
handheld device) fill all the space around with audible pattern of
sound. So, the sound entering my ears from the device can be
different and sound like its coming from you, while the sound entering
your ears from the device can be different and sound like its coming
from me.
The device can also identify your voice and correlate your voice with
your Ears. This would be only a slight improvement over existing AFR
software
http://vismod.media.mit.edu/vismod/demos/facerec/system.html
Basically the system would know you and know where your ears were
relative to your voicebox. And so, I could speak english and you
could speak Swahili and another person German and the system knowing
this from an RFID chip on our persons - would pick up my English and
your Swahili and the other person's German and translate things so it
sounded natural continuous and in each of our respective languages.
Also the people standing at the console looking at a display in
different languages, could also talk to the computer in different
languages and get responses back in their native languages - just as
easily as the signage provided them instructions in their language.
Software that could learn a language would be more sophisticated than
the software that translated and kept intonation and pitch and so
forth consistent across all languages - so, recognizing and
translating existing languages would come before the ability to pick
up and learn other languages.
But that may be possible too.
By 2040 some experts predict that human level computing will be
commmonplace. Well before that we'll have computer software that
exhibits common sense and understands reality. Virtual reality
systems will have this. Gaming systems today incorporate a lot of
realism by understanding in the context of the game - how physical
reality works. Operating systems that incorporate this knowledge at
all levels allow a more intuitive approach to computer interaction -
and provide a context for computer or human interaction
http://people.msoe.edu/~welch/courses/cs384/papers2004/stapelbroek.pdf
Put differently, two people sitting in a ship's cabin interacting with
a one another while a translator is present, would have a context for
their interaction by being in that environment, and the environment
would be easily modelled by the computer. So, this would give a
context to all interactions and sound patterns.
The first thing would be to notice that when someone said something in
a known language, someone would respond a certain way in an unknown
langage. From this pattern and the context of the interaction -
including reading body language and perhaps even medical data of both
parties - psychological and emotional states can be guessed at and
correlated with sound patterns, body movement (for sign language) and
so forth.
http://www.computerworld.com/softwaretopics/software/story/0,10801,108320,00.html
In a world of increasing mobility I can imagine developing computers,
entertainments, and signage that is able to respond in a language
appropriate way to people walking by and using it on demand.
A more sophisticated system would translate conversations on
telephones and video phones and teleconferences and the like.
More sophisticated still are dynamic interpersonal translations among
speakers of known langauges.
More sophisticated still are adaptive systems that can learn new
dialects knowing existing languages
ANd finally adaptive systems that through reading context, body
langauge, medical data and brain pattern data - can learn entirely new
languages and translate them dynamically on an interpersonal level.
Any communications systems, be it portable communicators, telepohones,
video phones, squawk boxes adn the like, could be equipped with
software that implements these levels of capability - allowing
seamless communications among a number of diverse users.
http://en.wikipedia.org/wiki/Phased_array
http://www.phased-array.com/1996-Book-Chapter.html
http://www.srh.noaa.gov/srh/jetstream/remote/remote_intro.htm
http://www.brooksidepress.org/Products/Military_OBGYN/Ultrasound/basic_ultrasound.htm
Another interesting point is that technological advance promotes
increases in the rate of technological advance. This means that an
infinite amount of technical advance will occur in a finite time.
That is, about 2040 AD - many predict a TECHNOLOGICAL SINGULARITY to
occur - which will result in ALL advances being completed by that
date.
http://en.wikipedia.org/wiki/Technological_singularity
I'll be 87 - and hopefully that will include control of the peculiar
degenerative process we call aging! haha..- and we should also have
universal translators as well.
So there you go.
.
- Prev by Date: Re: B5: Lost Tales
- Next by Date: Bread and Circuses - Plot, Effects, Cut Scenes
- Previous by thread: B5: Lost Tales
- Next by thread: Bread and Circuses - Plot, Effects, Cut Scenes
- Index(es):
Relevant Pages
|
