[FoRK] Voice Recognition (Training) Profiles

Reza B'Far reza
Fri Oct 28 08:44:43 PDT 2005


We're working on leveraging existing VR systems for speech transcription
(large vocabulary recognition).  What you're suggesting is not possible for
this.  Basically, there is no standard approach right now to speech
transcription.  The first step would be to come up with a standard approach
(which is not currently possible because the no one has invented/discovered
the canonical mathematical model).

However, your suggestion IS possible for so called "Command-and-Control"
(small vocabulary recognition used for controlling an interface -- all you
need is the basic commands like "Next","Back", etc.).  I don't know of any
efforts here, though if you try to use Sphinx, Nuance, Scansoft, or IBM,
you'll basically find out that their systems are so similar that, from the
engineering point, creating a profile like you're suggesting is a very
doable effort... would just require someone who designs a nice
Machine-to-Machine interface (Web Service, etc.) and keeps a registry of
users audio profiles.  The questions that come to mind at that point are
economic:

1) would users be willing to pay for this so that if you created something
like this (which would probably be costly in terms of storage and bandwidth)
you could at least break even
2) would third party vendors cooperate and integrate

To my experience, the second is probably a "No".  For whatever reason, VR
ASP's in the voice world are non-cooperative... probably because the market
is still so small that they cannibalize one another to continue existing...
or may be that they're just all greedy... I am just guessing on this one...

Reza B'Far, CTO
Voice Genesis Inc.

-----Original Message-----
From: fork-bounces at xent.com [mailto:fork-bounces at xent.com]On Behalf Of
B.K. DeLong
Sent: Friday, October 28, 2005 5:48 AM
To: fork at xent.com
Subject: [FoRK] Voice Recognition (Training) Profiles


If anyone would know about the development of such technology, it
would be FoRKers. Anyone heard of such developments?

See:

http://www.livejournal.com/users/bkdelong/139700.html

--
B.K. DeLong
bkdelong at pobox.com
+1.617.797.8471 (Note new number)

http://www.brain-stream.com               Play.
http://www.bostonredcross.org            Volunteer.
http://www.the-leaky-cauldron.org        Potter.
http://www.hackerfoundation.org          Future.
http://www.wkdelong.org           Son.


PGP Fingerprint:
38D4 D4D4 5819 8667 DFD5  A62D AF61 15FF 297D 67FE

FOAF:
http://foaf.brain-stream.org

_______________________________________________
FoRK mailing list
http://xent.com/mailman/listinfo/fork




More information about the FoRK mailing list