Windows 7 Win 7 voice recognition question

grindy

Senior Member
Joined
Jun 26, 2009
I have literally hundreds of hours of excellent quality audio of my (late) father speaking. I would very much like to be able to use Windows 7 voice recognition to transcribe these audio's into text, but from what I know about speech recognition apps - they all seem to require training via a microphone (not an audio recording). My question is whether any of you here have managed to do what I would like to do, either with Windows 7 or any other voice recognition app?
 
Win7 Speech Recognition is primarily for voice commands to the PC.
But, apparently it can be used for dictation.

It doesn't look like you will be able to just play your father's recordings to get the text.
There is a learning period where the app learns the person's voice and speaking patterns.
Then, you'll have to add punctuation [period, stop, comma, etc] by speaking the punctuation where it is needed.
And finally, there is little doubt that it will take a lot of editing to get everything right.

Even Dragon Naturally Speaking requires a learning period.
I don't think the Windows Speech Recognition will work as well as Dragon does.
I think you are asking for more than normal run of the mill reasonably priced to free software can do.
 
Last edited:
I agree with Voyager in that is will be very hard, if it will work at all. Dragon Naturally Speaking is probably your best bet, but have you tried anything so far?

How is the audio recorded, and have you thought about its long term storage?
 
@ Voyager: Thanks for responding... Please note that in my original post I mentioned Windows 7 or "any other" voice recognition, and don't expect W7's to adapt...
@ Saltgrass: Dragon Naturally Speaking seems to require the same microphone training that W7 does... The audio is all mp3, excellent quality (recorded with a professional mic)..

It seems to me that the ability to use a speech recording is something that reverse training should be able to do. By that, I mean to somehow give it - let's say an hour - of recorded speech along with the written text of exactly what was spoken in that speech, and train it that way... But, it's not looking good so far... I think the purveyors of Voice Recognition software are missing a big market opportunity here...
 
There may be something out there that can do what you want.
But, It'll probably be in the kilo-buck or higher range, something governments or enterprises might be willing to pay for.
You are asking for a very sophisticated type of software.
 
Last edited:
There may be something out there that can do what you want.
But, It'll probably be in the kilo-buck or higher range, something governments or enterprises might be willing to pay for.
You are asking for a very sophisticated type of software.

I suspect you're right, voyager...
 
Back
Top Bottom