Re: [ARSCLIST] voice recognition software

[Christina Hostetter <chostetter@xxxxxxxxx>]

Ironically, this exact topic came up on the Archives and Archivists
listserv this afternoon.  Here is one of the responses to that question:

"I've used Dragon for transcription of my own talks and for dictation.
My experience is that it has worked very well for that.

I'm sorry to say, it would not work well for oral history transcription
since the program needs to be "trained" to recognize the speaker's
voice. Training is handled by reading various passages into the program
so that it learns to recognize your voice."

 This has been the general consensus on the matter.  We talked about
doing that at our institution and decided against it.

~~~~~~~~~~~~~~~~~~~~~~~~~
Christina J. Hostetter, MA, MLS, CA
Archivist
Eric Friedheim Library at the
National Press Club
529 14th Street, NW
Washington, DC  20045
Tel: (202) 662-7523 
Fax: (202) 879-6725
www.press.org/library/archives 
~~~~~~~~~~~~~~~~~~~~~~~~~


-----Original Message-----
From: Association for Recorded Sound Discussion List
[mailto:ARSCLIST@xxxxxxx] On Behalf Of Brandon Burke
Sent: Thursday, December 15, 2005 4:33 PM
To: ARSCLIST@xxxxxxxxxxxxxxxx
Subject: [ARSCLIST] voice recognition software

Friends,

It has been suggested that our institution purchase voice recognition
software for the transcription of oral history interviews.  Naturally, I
am against this decision.  The product in question is Dragon Naturally
Speaking but I imagine others would satisfy this person equally.

as I see it...

* Digital recording devices need to be "trained" before every session.
That means interviewees have to sit there and read the phone book out
loud for several minutes.  I can't imagine that going over well..

* It's probably difficult for VRS to understand what to do when
multiple, concurrent accents.  (Ex: person #1 is from Texas but #2 is
from France.)

* VRS probably reacts unfavorably to loud background noise.

* Punctuation likely require serious editing, enough so to question why
the software "saves time".


Arguments either for and against VRS are welcome as are suggestions 
regarding improvements to the classic cassette-recorder situation.  I 
should tell you as well that I'm thinking about moving up to a hand-held

digital recorder.  Something like the Marantz PMD-600 series...namely 
because they tend to have USB outs.

thanks as always,
Brandon


_______________________________________
Brandon Burke
Assistant Archivist for Audiovisual Collections
Hoover Institution Archives
Stanford University
Stanford, CA 94305-6010
voice: 650.724.9711
fax: 650.725.3445
email: burke@xxxxxxxxxxxxxxxxxxx