User records a Pass phrase of his/her choice (one or more times) on the device.
After the training, he/she has to speak the same phrase in order for the system to
recognize him/her and allow access.
If a person, other than user speaks the same phrase, the system rejects him/her with
reasonable degree of accuracy.
There needs to be some noise immunity in the system in the sense that user might have
trained the system in quiet environment but might want to use it in some noisy environment
so the system needs to work.
If possible, we'd like to use pocketsphinx like front end for VAD (voice activity detection)
and feature vector creation.
Is such a thing possible with any of the Sphinx versions available ? or could somebody please
direct me to some papers/references which can be of help for such implementation.
Thnx,
Li
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi,
We are trying to solve the following problem:
recognize him/her and allow access.
reasonable degree of accuracy.
trained the system in quiet environment but might want to use it in some noisy environment
so the system needs to work.
and feature vector creation.
Is such a thing possible with any of the Sphinx versions available ? or could somebody please
direct me to some papers/references which can be of help for such implementation.
Thnx,
Li
No, this is not available out-of-box
Utterance verification On DTW based speech recognition using likelihood
http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=5620715