Transcription
I ran the audio of my Library of Congress presentation through DragonSoft, thinking that this would be an easy way to make a transcript. Here’s how it renders the opening:
To him here are sure you are in the OK when loan. And I have a want a and were a Shannon I. here in him that we here, third and or Shannon, there will you a very moral him the now I will to him eight. All for your has no longer so I know I’ll tell you are you sure you will I know him the eight him she and her to him we will be in the only to the him to all or me one alone are sure to in yeah, I here you to him are the only in normal, her know there are no more I mourn air him sure all the handle or in yeah, only yeah OK I did yeah for your third him I will or a travel and hey him in I mourn room will sure you sure you a little. The interim, hey oh, that her him to enter yeah there will all or of the and they yeah, yeah yeah fo their him a little room should him in him in him the him. We him to you or you all or yeah r yeah
I seem to have the “Translate into Molly’s Sililoquy” option on.
Yes, I did train the software. In fact, if I play the WAV file over the speakers into a microphone, DragonSoft does much better…but not good enough to make it an appealing option for this project.
Categories: Uncategorized dw
test
Have you tried reading the above back into Dragonsoft? Maybe it will translate back into the original … but seriously, does DS work under any practical circumstances? I’ve been tempted to get it but decided it would prob never work well enough (although I believe its getting better).
Yes, DragonSoft works pretty well under normal dictation circumstances. I still type faster and more accurately than it does so I’m not tempted to throw away my keyboard. But I’ve seen it work well for the dyslexic, for example.
one wonders what’s diffrent about the library of congress recording versus ‘ideal circumstances.’ Perhaps it would be possible to tweek the recording to better match them? Like the C-Span microphones.. (or whoevers) one would guess are a good deal better then the one’s one normally uses with dragon.. perhaps fooling around with a multi-band compressor and parametric EQ might help it? Not that it would be worth it.. Though maybe if you figured this out it might be worth it in the log run.. and anyway, doesn’t C-Span ussually offer transcripts anyway?
insidently.. Paul.. a guy I used to work with at the EBX in natick swairs by Dragon.. (This for the benifit of Vergil)