Dear Marten,
Using voice recognition for transcribing interviews is not as simple
as it may appear to be at first sight.
I transcribed sixteen qualitative interviews totalling 135 thousand
words from tapes. The research was for my PhD. Because I had
conducted the interviews I wanted to transcribe the tapes myself so
that I could include annotations in the text such as hesitation in
speech, laughter, groans etc. I also wanted to tie the transcripts in
with the written notes I'd taken at the time of the interviews. These
included notes on interviewees facial expressions and gestures.
I used a transcribing machine with foot peddle and purchased
ViaVoice software. However, speech recognition software has to
be trained to the user's voice. Therefore it cannot be used to
transcribe what is said by the interviewee.
After training the software to recognise my voice I listened to the
tape through headphones and repeated sentences into the
microphone. The process was far from easy. The microphone had
to be repeatedly switched on and off because if it was left on whilst
the tape was playing it picked up the interviewee's voice and
interpreted it as garbage. As with any transcription I frequently had
to reverse the tape to confirm what had been said. The combined
tasks of forwarding and reversing the foot peddle, listening to the
tape, switching the microphone on and off and repeating sentences
is extremely time consuming.
I struggled on through about two or three interviews and then
abandoned the process in favour of straightforward listening and
keying in. As I have keyboard skills this was by far the easier
option. In actual fact I wasted weeks trying to use voice
recognition to transcribe the tapes. I was convinced that it would
work if I perserved, but I was wrong.
Just thought my experiences may be of use. This was in 1998 and
voice recognistion software may have improved since then. Others
may have had better experiences but the soft option I thought I had
found just didn't work out for me.
Heather
> Hello to all,
>
> I am wondering if someone knows anything about the use of voice
> recognition and conversion of recorded interviews to text. It should
> save a researcher lots of time, if she or he wouldn't have to write
> out interviews manually. I have never heard of people using voice
> recognition in qual analysis, but i reckon it must be possible -- the
> use of it in telecommunication is quite advanced.
>
> I hope someone can help me
>
> Greetings,
> Marten
>
>
>
>
> ~~~~~~~~~~~~~~~~
> Drs. M. van der Meulen
> PhD-student Sociology of Religion
> Faculty of Theology
> Vrije Universiteit (Free University) Amsterdam
> De Boelelaan 1105
> 1081 HV Amsterdam
>
> +0031 20 4446631
> [log in to unmask]
> www.th.vu.nl/~marten
>
****************************************
Heather Williams
Research Fellow
School of Environment and Life Sciences
University of Salford
Allerton Building
Frederick Road
Salford. M6 6PU
United Kingdom
Telephone: +44 (0)161 295 3155
E-mail: [log in to unmask]
****************************************
|