A question commonly posed by social scientists undertaking interviews was raised in the Ask Jack column on the Technology pages of the Guardian last week: “I am about to start a university research project involving in-depth interviews. I am aiming to download audio files of interviews to my computer from a digital audio recorder. I would then like to use software to turn the recorded voices into text” . The familiar response was: “you can’t do it”. Although speech-recognition software like Dragon Naturally Speaking achieves impressive accuracy when trained to recognise a particular user’s voice, it cannot capture a host of different voices from research interviews. Human transcription is needed, and Jack Schofield offers advice about using a good quality digital recorder, a separate microphone, and transcriptions software with a USB foot-pedal.
Commentators on his blog add further suggestions. There is also a Social Research Update on ‘Tools for Digital Audio Recording in Qualitative Research’ written by Alan Stockdale and published by the Department of Sociology at the University of Surrey in 2002 which contains some useful advice. Even though transcribing interviews is a major hurdle in qualitative research, regrettably it is unlikely that the application of High Performance Computing or other e-Science resources will offer a solution even in the medium term. Converting numerous conversing voices into quality text is too complicated. This is one area where NCeSS cannot assist.