[Air-L] out of luck re transcription?

scott at scottmacleod.com scott at scottmacleod.com
Wed Jan 28 11:17:13 PST 2009


Here are CNET reviews of Dragon Naturally Speaking 10 from August
2008, and another one of version 8 from 2006. 

The most recent review says DNS 10 gets about 4 out of 5 words
correctly, and that
refers to one's own voice, not other people's voices. 

I think this may be the best transcription software available, and
it's seems like it's still not ideal for interviews. Check the web for
other reviews. 

Scott
http://scottmacleod.com 


CNET review of Dragon Naturally Speaking 10 from August 2008
http://reviews.cnet.com/voice-recognition/dragon-naturally-speaking-10/4505- 
3528_7-33227363.html?tag=mncol;lst 


And here's CNET review of Dragon Naturally Speaking 8 from 2006
http://reviews.cnet.com/voice-recognition/dragon-naturallyspeaking-professio 
nal-8/4505-3528_7-31228939.html 


T. Kennedy writes: 

> We used Express Scribe - Meryl already mentioned it - and it works great.
> I'm not aware of any programs that transcribe direct successfully.
> I have to agree with Denise, it's certainly part of the process (as painful
> and tedious as it is) - and you get much closer to your data (and
> importantly, the participants). Having said that, 50-75 interviews is a huge
> undertaking for a single grad student (who is not part of a research group)
> - even for a phd.
> t 
> 
> Tracy L. M. Kennedy
> PhD Candidate
> Dept of Sociology
> University of Toronto 
> 
> Course Instructor
> Dept of Communications, Popular Culture & Film 
> Brock University 
> 
> Research Coordinator
> NetLab
> University of Toronto 
> 
> Second Life: Professor Tracy 
> 
> -----Original Message-----
> From: Barry Wellman [mailto:wellman at chass.utoronto.ca] 
> Sent: January 27, 2009 4:14 PM
> To: aoir list
> Cc: Tracy Kennedy
> Subject: out of luck re transcription? 
> 
> We've never found a transcription program that will work with interviews.
> That's because you need to train the program for each voice, and while the
> interviewers are trainable [;-)], each respondent isn't. Hard enuf to get
> a resp. to hold still to be interviewed. 
> 
> We have had good luck with transcription software, for going from voice
> recording to interviews. Tracy, what did we use last? 
> 
>  Barry Wellman 
> 
> 
> _______________________________________________
> The Air-L at listserv.aoir.org mailing list
> is provided by the Association of Internet Researchers http://aoir.org
> Subscribe, change options or unsubscribe at: http://listserv.aoir.org/listinfo.cgi/air-l-aoir.org 
> 
> Join the Association of Internet Researchers:
> http://www.aoir.org/
 




More information about the Air-L mailing list