Windows hello application based on voice print recognition
Abstract
Voiceprint recognition is a special category in speech recognition. It can be
commonly called speaker recognition. In recent years, with the artificial intelligence
gradually entering the public eye, the authentication method based on the neural
network has been further developed. For example, face recognition, iris recognition,
and voiceprint recognition. These identity authentication methods gradually evolved
from the theoretical stage into practical applications. For example, Apple's Face ID –
Face Recognition and Recognition feature released at the end of 2017, which
improves the way traditional password authentication and your fingerprint
authentication can quickly unlock your iPhone without having to remember any
combination of passwords. There is also the Windows Hello authentication
application, for example, that Microsoft proposed in 2015; it uses the identity
information collection device provided by Microsoft to record the information of the
pre-landing person into the system, and uses the embedded algorithm to perform the
judgment of the pre-landing person, thereby realizing Log in quickly and start
working.
With the development of technology, identity authentication technology
continues to simplify the authentication process, optimize the authentication
efficiency, provide protection for information security, and increase convenience for
identity authentication.
In this project, Gaussian mixture model is used to extract the speech signals
collected from the microphone for MFCC diagnosis parameters and put into the
model for training. The matching degree between the pre-landing person and the
person in the speech library is calculated to realize the identity authentication
评论0