Last post Apr 19, 2015 07:30 AM by Rion Williams
Apr 17, 2015 03:06 AM|xyz1511|LINK
Can I Use Dictation For Converting File Audio wav (Like Conference Recording) To Text?
I am using Microsoft.Speech for converting Audio file to Text , is there anyway to use the speech recognition to listen to what a user says from audio file without having to load the choices in a grammar object? For example if you ask the user for their
name, it would be tough to find all options a user can give but it would be nice if it could give you a result and the confidence on the result. Or is this not what the speech recognition is intended to be used for? Thanks.
Apr 19, 2015 07:30 AM|Rion Williams|LINK
As I am sure you have come across
this related discussion based on your question as it pertains to the built-in Microsoft.Speech library, which has a few ideas on how to accomplish this through the use of including built-in dictation grammars using the Microsoft.Speech.Recognition namespace.
However I personally cannot vouch for it or provide any kind of confidence with regards to its accuracy.
In my opinion, if you want any type of production-level accuracy, you would likely need to use some type of third-party software or service to handle transcription :
As you can see there are quite a few APIs out there that you may be able to take advantage of. Additionally, it should be noted that if you are handling any type of critical data (i.e. medical records, financial data, etc.) that you should probably have
a stringent QA process to evaluate the accuracy of the conversions or perhaps use a professional transcriptionist to handle this process.