 |
 |
I think its highly unlikely that we shall see voice recognition software that will handle "conversation speech" in the next decade. The central issue, apart from recording quality, is that when people converse they don't follow the rules of natural speech and the software uses these rules to decide between different possible words. Using voice recognition software as an intermediary to transfer the text is one solution but it has problems in terms of the accuracy of the speech, for instance ums and ahs are difficult to reproduce. My own experience using Via Voice is that it works very well if I am transferring formal text (the vocal equivalent of copy typing) but not for conversational speech. For the latter, the amount of time spent making the corrections to get the text accurate is longer than it would take to transcribe direct. Code-A-Text offers another solution. You transfer the audio file to a digitised format (the facilities are part of the programme) and then code against the sound file itself. You thus always have access to the original recording (you can do the same with video). You can then either write precis of each segment or just transcribe those sections which are important. Regards Alan Cartwright At 07:55 03/02/99 -0600, you wrote: >If only there were such a thing as the capability for quality, direct >transcription of audiorecordings! Dragon software is excellent for >general voice purposes. The user could listen to the tape, then speak >the content into the program with good results, but this is not exactly >direct transcription. There is a program that requires less stringent >user profiles, though, meaning it does not have to be "trained" for a >specific speaker. I *think* it is Via Voice, which is able to recognize >speech with only a gender profile (separate profiles for male and female, >regardless of specific speaker). I think the quality of the recording is a >big issue here, though. The accuracy of voice recognition software >depends, to a great extent, on having a quality microphone, properly >placed to pick up the input. How to transfer that process to a tape recording >is beyond me. I have tried hooking up a tape recorder directly to the >mic. input, but the quality was sorely lacking. I think it is only a matter of >time before this becomes a reality, however. If only the manufacturers >knew what a market there was! > > >Beth L. Rodgers, Ph.D., RN Tel: 414/229-5466 >Associate Professor Fax: 414/229-3987 >University of WI--Milwaukee Web: http://www.uwm.edu/~brodg >School of Nursing Personal: [log in to unmask] >Email: [log in to unmask] > Alan Cartwright PhD Code-A-Text Developer Email [log in to unmask] web page http://www.codeatext.u-net.com Also Senior Lecturer In Psychotherapy Kent Institute of Medicine and Health Sciences. University of Kent. UK. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%