Title kinda says it all. I have a friend who does interviews but hates transcribing the interview off of tape. I was wondering if there was dictation software (Dragon or Vista’s built in version or something else) that can take that audio file and transcribe it?
I doubt it would work as the interviewed person has not sat down and trained the software how they pronounce words. Word recognition would be bad.
I was reading a review of Dragon Naturally Speaking software and it said it did a pretty good job with no training (although training improves things).
In the end I will test and assess the performance. Just cannot tell if I can just throw and audio file at it and have it translate that. I cannot think why it sholdn’t be able to do that but who knows?
Free wave to text software. I don’t vouch for this particular software. I list it as an example of what a search turned up. It’s entirely up to you to test this or find something different.
As far as I’ve heard, DNS is a pretty good product. If it had a free trial, I’d say; just check it out, but as far as I can find, it doesn’t. Still, if you can afford to spend $50 dollars on something that may not work, you could consider it.
Hmm…I checked the Dragon features matrix and the cheapest version that seems to allow transcription from an audio file is $199.
I’d really prefer to try it first but I suspect they have learned people who try it are always a little let down that it is not Star Trek capable voice commands and give it a pass.
You don’t have Vista or Windows 7 installed? You could try just plugging a voice recorder in the line input and test it on someone who does have it and see if that works well.
Otherwise, I just found this has a free trial: http://www.research-lab.com/vexp007read.htm No guarantees.
Dragon has a FREE dictation to text app for the iPhone. Works remarkably well with no training. If their current PC software is anywhere near as good I;d say to give it a try.
There are just too many variables to consistently get an untrained voice recognition program to work well. I’ve done it, and even with crystal clear audio, there will be a spectacular number of atrocities in the finished work. Add multiple voices, as in the case of an interview, and it gets even worse.
The best way to do it is train the program - I use Dragon - and then listen to the audio in headphones/earbuds/whatever while you repeat what you hear into Dragon. It’s really not as awkward as it sounds. This way, you can also use the other voice commands via Dragon to add non-spoken information to the text as you go along.
If I have a case to do where everyone speaks clearly and there’s no background noise, it’s still faster for me to just type everything. However, when I have stutterers/stammerers or people with thick accents, Dragon works great because I can do the mild paraphrasing “internally” and say exactly what I want typed out.