I'm new to the speech recognition area. I tried to modify the pocketsphinx android demo program to do digit(password) speech recognition. I used the cmudict-en-us.dict for the dictinary and the digit.gram file for the grammar. Then it gave me really bad recognition results (e.g. When I speak a series of numbers it ignores the last one; The recognition correctness is really low; It recognizes noises as numbers frequently). The similar issue occurs when I use keywords to wake up my app. It recognizes noises as keywords frequently even if I already set the threshold to be "1e-1". What can I do to get rid of the issue? Thanks.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I'm new to the speech recognition area. I tried to modify the pocketsphinx android demo program to do digit(password) speech recognition. I used the cmudict-en-us.dict for the dictinary and the digit.gram file for the grammar. Then it gave me really bad recognition results (e.g. When I speak a series of numbers it ignores the last one; The recognition correctness is really low; It recognizes noises as numbers frequently). The similar issue occurs when I use keywords to wake up my app. It recognizes noises as keywords frequently even if I already set the threshold to be "1e-1". What can I do to get rid of the issue? Thanks.