I understand speech recognition for "Accept"/"Reject are very high.
I also understand when you train the Genie, the Genie first considers the trained recorded version of a command, if it fails to match then it consideres the original speech rec version of the command. (2 chances for success).

Would allowing the Accept/Reject to be trainable, while leaving the tolerances high on the original speech rec method improve the success of Accept/Reject?

I'm looking for higher acknowledgement success on hands-free operation. -- Thank you.