WebAIM - Web Accessibility In Mind

E-mail List Archives

Re: tool to convert oral recording to text?

for

From: John E Brandt
Date: Jun 20, 2011 12:09PM

At the Accessibility Unconference I attended last year there was a person
representing a company that did speech-to-text (STT) transcription of video
recordings. He claimed it was of fairly high quality but did indicate that
human editing was required to ensure accuracy. In the conversations that
followed, there was a general consensus that the technology to do accurate
automatic STT transcriptions from video is still not there.

Yes, Dragon Dictate does a reasonable job with one voice in a controlled
environment, but I don't think it can do multiple voices in the same setting
very well.

My suggestion would be to hire a good transcriptionist like a CART person if
you want an accurate STT transcript. The biggest advantage of the human
being is that the human can stop the proceedings and ask people to repeat
themselves when there is confusion or if multiple people are speaking.

You will recall that IBM's Big Blue genius "Watson" could not respond to
verbal content on Jeopardy....and IBM engineers were the ones who developed
STT.

John E. Brandt
www.jebswebs.com
<EMAIL REMOVED>
207-622-7937
Augusta, Maine, USA