Future System TABA
This research prototype, the automatic-telephone-railroad-information-service (TABA), is
currently running on an experimental basis.The automatic information service works on the basis
of a language-recognition-system and is familiar with train connections between more than 1,000
German destinations . TABA is capable of understanding questions formulated in fluent everyday
German. Each question is being digitalized and recorded. Afterwards the language-recognition-
system checks the text and computes a certain number of possible word sequences. Assisted by a
built-in grammar the dialogue module breaks down these sequences within a few milliseconds,
choosing the version that makes the most sense. In case any specific information needed to find
out the required train connection is missing, the system asks it of the caller. As soon as all
necessary information has been gathered, a number of alternative train connections are chosen
from the stored schedule and presented to the caller. Currently, three out of four calls are
recognized satisfactorily by the system. Since it receives a new sample of spoken language each
time it is called, TABA is improving constantly.
Speech Processing System 6000 (SP6000)
A dictating system, automatically transforming naturally spoken texts into writing. The system
comes with a built-in vocabulary of around 60,000 words and specific terms from radiology. For
frequently dictating practitioners, such as radiologists at hospitals or doctors' offices, this is a
future-oriented way of formulating diagnoses, increasing productivity and guaranteeing faster
diagnoses around the clock, even in emergency cases. The first two systems have already been
installed in hospitals.
Continuous Language-Recognition as a Modul Palette
The continuous language recognition system presented as product SP6000 is offered in form of
diverse modules. The first Beta-versions have already been delivered to some VARs. The
language-recognition modul-palette includes the dictating module SpeechNote, the language-
recognition module for continuous language-recognition SpeechMagic, and the context-
development-tool SpeechPro. What can be presented under these circumstance are ways of
recognizing a "futuristic" office languag. So far the "computerese" is available in English only.
Voice System 2000
VS2000 is a digital dictating machine, developed for smaller organizational or user-groups with
centrally organized dictation. Due to a "Plug and Play"-installation the dictating system is ready
for immediate use. Small and mediumsized enterprises, such as lawyers' or doctors' offices (with
various authors dictating and 1-2 employees writing), will find VS2000 a communication
instrument that makes possible cost-conscious and effective working. The dictating system is in in
use for 24 hours and can be reached worldwide from any digital telephone.
| |
Philips Dictation Systems, Vienna
