From: "E.J." <edseljoe@earthlink.net>
To: "USF Anthropology list" <usf-ant@lists.cas.usf.edu>
Sent: Saturday, August 28, 2004 9:24 AM
Subject: RE: An end to transcription
The problem with a lot of those audio transcription programs like Naturally
Speaking, is that they generally have to be trained to transcribe each
individual speaker. The programs have to be trained to recognize speakers,
typically by giving them a word-list of some kind which the program then
uses to interpret the noises coming out of your mouth.
Although these programs work pretty well for one person, they work poorly
when more than one person (especially if one of the persons has a very
unusual pronunciation) is speaking. I suppose you could give the word list
to every interviewee so that later you could train the computer to recognize
that person, but you would effectively have to multitrack the transcription
process, running the transcription through once for the first speaker, then
retraining the recognition software and re-running the program for the
second.
Heaven help you if you have more than two speakers!
EJ "Yes, I am a Luddite" Ford
-----Original Message-----
From: bounce-usf-ant-16236@lists.cas.usf.edu
[mailto:bounce-usf-ant-16236@lists.cas.usf.edu] On Behalf Of Caroline
Peterson
Sent: Friday, August 27, 2004 6:24 PM
To: USF Anthropology list
Subject: An end to transcription
There are options to old fashioned transcription. Programs exist
that are voice activated transcription. Just read/speak into a
microphone and the program types out what is said. This can be used
for interviews or for inputting data etc.
One program with which I'm familiar is cheap, easy to voice train,
and really accurate. I saw it at a chiropractic conference and they
said it has pre-programed terms for different disciplines including
anthropology.
Take a look:
http://www.NaturallySpeaking Direct.com/
Voice Factor/PCS, Inc.
15225 US Hwy 19N., Ste 202
Hudson, FL 34667
727-697-2800
Check it
