My wife, for a linguistics research project, recorded the responses of her
subjects to a list of questions, so that she could transcribe them and
analyze them. She got Dragon Dictate with the intention of getting a rough
transcription and then, listening to the tape, correct the errors. With so
many subjects, she found that the software, which was trained to her voice,
never gave good enough results to be faster than transcribing them herself
by hand. So, keep that issue in mind. I would imagine that, if there were
just a few lecturers for the podcast, that a speech transcription solution
might work out, provided you could get the lecturers to spend a little time
helping to train the software. However, you need to find out how many
different voices a given dictation software can be trained to reliably
understand. If it is fewer than the number of lecturers, it probably not be
worth it. If it did work, you could send rough transcriptions and the audio
and have Mechanical Turk do clean up edits rather than the whole
On Mon, Feb 27, 2012 at 1:52 PM, Suchy, Daniel <[log in to unmask]> wrote:
> Hello all,
> At my campus we offer podcasts of course lectures, recorded in class and
> then delivered via iTunes and as a plain Mp3 download (
> http://podcast.ucsd.edu). I have the new responsibility of figuring out
> how to transcribe text versions of these audio podcasts for folks with
> hearing issues.
> I was wondering if any of you are using or have played with
> dictation/transcription software and can recommend or de-recommend any?
> My first inclination is to go with open-source, but I'm open to anything
> that works well and can scale to handle hundreds of courses.
> Thanks in advance!
> Daniel Suchy
> User Services Technology Analyst
> University of California, San Diego Libraries
> [log in to unmask]<mailto:[log in to unmask]>