hello all,

i'm looking for advice. i require a speech to text software which would take audio files from disk from a number of different speakers and its result would be a transcript of the audio file in question. i'd love for this to be batchable as well. accuracy is a key, punctation is note.

if anyone knows a piece of software able to do this kind of work. please let me know.

thanks
thekenny

Hi Kenny,

John Wickett recently answered this question on our forum. (we have received numerous request for this technology) I have inserted his remarks below. Thank you, hope this information helps...

"Unfortunately, voice recognition programs need to be trained to the individual users voice. It is not yet at the point where one voice file will work with multiple users. However, there are a lot of companies working to try to make it happen, so I don't think you'll have to wait too long."

John
Dictated with NaturallySpeakingjwickett@auracom.com
jcwickett@vac-acc.gc.ca

I wanted to ask, just how the estimates are about that universal recognition, or better, what is the accuracy today.
Or could you point out the firms that are dealing with this issue?
thanks.

Hi Genosid,

The Accuracy of Speech Recognition Software depends on the product and the hardware. There are several companies that have developed speech recognition engines; to date Dragon NaturallySpeaking features higher accuracy and the best overall performance.

If installed on proper hardware you can achieve 98% accuracy consistently.

Thanks for participating in the forum,


Krish Harmeier
Technical Director
21st Century Eloquence
Voice Recognition Specialist
1.800.245.2133 ext. 101
krishh@voicerecognition.com