[FoRK] BrainStream: Subvocally-controlled,
augmented-reality geospatial interactions
<bkdelong at pobox.com> on
Fri Sep 7 13:05:25 PDT 2007
I got hit with a BrainStream around lunch today this fine Fri afternoon.
Would love comments, thoughts, contributions, suggestions and any
corrections. Often times my mind goes for a loop and gets lost in
fantasy and impracticality but I can always count on some good input
from FoRK and folks in various futurist and transhumanist communities.
I've been waiting to write about this all morning (via BoingBoing). It
seems a bunch of intrepid researchers at the Rehabilitation Institute
of Chicago have taken work done on subvocal recognition (SVR) and
applied it in such a fashion that one can drive/control a wheelchair
through "thinking" about where they want it to move.
(NB: I've contributed heavily to the SVR wikipedia entry and liberally
link to said service below. I acknowledge that most entries are not
that of expert research so YMMV.)
I first got excited about SVR after reading Cory Doctorow's "Down and
Out in the Magic Kingdom" in which the protagonist makes phone calls
and interacts with his PDA subvocally. That is to say, the device
detects electric signals sent from his brain to his larynx which would
normally be translated into speech in the vocal tract. NASA and DARPA
both had done some work in the past few years through the subject
wearing a collar which detects the signals.
I think my endgame with SVR is really what Cory had in mind in "Down
and Out" - be able to not only talk silently on the phone but to
control a PDA, surf the web etc.
The phone hardware or service would convert the signals to speech -
think of it; ring-tones now and in the future voice profiles - make
your voice sound like anyone in any language. Somehow I don't think
the Intelligence Community would appreciate that. It makes it
difficult to determine who is who on the phone and would kill Voice
Stress Analysis. I'm guessing the FCC and/or the ITU would want to
have some say in how that works. Other issues include protecting ones
own voice profile and the damage this could do to voice-based
I wouldn't worry - voice recognition as it stands today is still
complete rubbish (kestrell any thoughts on that from the accessibility
side of things?). I still think the need to continually train any new
piece of voice recognition software is a waste of time and they should
all be required to adhere to a voice profile standard. Said standard
would, in theory, allow you to train something once and export to an
open format importing it into the same software (got to love the lack
of portability of licenses!) or any other similar products. At a
minimum, it should at least bring the user half-way to training the
program to work well. The big risk again is someone stealing this
profile to break voice-based biometrics. Surely that can be mitigated
with some nifty encrypted token of some-sort with decent key recovery.
There has been some interesting work done on voice translation
technology for the soldiers in the streets of Iraq that would have
commercial applications in the future. Think of babelfish meets voice
recognition and be able to speak with anyone regardless of their
native language. Can you say "Universal Translator" ?
But back to the sci-fi. Combine subvocal recognition with augmented
reality, GPS and gesture recognition (think Tom Cruise's manipulation
of a multi-input virtual screen in "Minority Report") tied into a
Net-connected PDA with a Head-Mounted Display (HMD). Hot damn.
The SVR would allow you to interact with the PDA silently. The GPS
combined with augmented reality as visualized through the PDA's HMD
would allow the user to view the entire physical world as annotated by
Semantic Web, metadata addicts similar to some of the art
installations touched upon in William Gibson's "Spook Country"
(current reading). The gesture recognition which would be determined
through sensors on clothes as well as the HMD would allow one to
interact with the augmented reality to manipulate the objects only
existent in the geotagged cyberspace environment. Absolutely
Ye gods I love taking lunch to paddle down a brain-stream.
B.K. DeLong (K3GRN)
bkdelong at pobox.com
38D4 D4D4 5819 8667 DFD5 A62D AF61 15FF 297D 67FE
More information about the FoRK