|
Speech Recognition
WinScribe dictation systems have the power to integrate Dragon
Speech recognition. Speech recognition is the computerized process of
changing speech into text. Currently WinScribe uses what is known as "back-end speech
recognition" as well as "front-end speech recognition".
Back-end SR is very easy to implement because it
doesn't require authors to change the way they dictate, many times authors
don't even know that speech recognition is in use. Telephone
users can continue to call in, PDA users can keep using Pocket Author or
Pocket Author for Healthcare, handheld users can continue using their
handheld and importer, and PC users can keep using WinScribe
Author. The speech recognition process takes place on a speech recognition server (a
dedicated machine) running the WinScribe speech software. When the
transcriptionist selects a voice file in Internet Typist, they receive a
copy of the recognized text. Transcriptionists now have the ability to
play the synchronized audio and text, make corrects and submit the document
for a process called adaptation. Adaptation occurs on the server and
is the process of reconciling the recognized text to the corrected
document. Adaptation makes adjustments to the individual author's
recognition profile and ensures that the system learns how the author speaks
becomes more accurate as time goes on. Over the course of time,
authors can experience over 99% accuracy with speech recognition.
Front-end speech recognition works well in many speciality areas where
authors are tied to a PC and many standard reports are used or a lot of
dictation is done. This does require authors to interact directly with
the software, so this is sometimes harder to phase in than back-end speech
recognition. Authors have the choice to correct the job themselves, or
send it on for editing by the typist.
|
Content reproduced
with permission from WinScribe ©
|
|
|
|
© Nelson Systems, Inc - All Rights
Reserved |
|
