Dec 29 2003, 06:42 PM
I have like 70 wav files that I recorded from a radio show, this show is only one person talking for approximatelly 40 to 50 minutes and I would like to transcribe these audio files. Is it possible to train DNS with these files?, I don't care if I have to transcribe two or three of these files manually.
Jan 9 2004, 12:46 AM
Yes it is possible and the details were posted here on the forum early last spring or summer, as I recall. The only caveat is that no words are allowed which aren't in DNS's built-in dictionary, so you'd need to take that into account, maybe by deleting the offending word(s) from the training WAV file with an audio editing program like CoolEdit.
After the training is done you can then add the new words to the dictionary during a standard proofreading/correction step, but not during the user training step.
[This message has been edited by Tom S (edited 01-09-2004).]