Probability linear speaker-distinguishing identifying method based on priori knowledge structured covariance

Abstract

The invention discloses a probability linear speaker-distinguishing identifying method based on priori knowledge structured covariance, which is capable of structuring a covariance hypothesis and iterative process of a probability linear identifying-analyzing model based on random useful information related to training voice; and finally, a probability linear identifying-analyzing model that can be more distinctive and can reflect the real situation can be trained; and at the same time, two structuring coefficients are introduced to make the model adjustable and can be self-adaptive to be optimum aiming to various different structuring information. By adopting the model trained by this method provided herein, compared with the traditional model, the evaluating effect of identifying the speaker on the same dataset is improved significantly; and the equal error rate (EER) and the minimum detect error cost function (norm minDCF) can be lowered 10 percent to 20 percent in the evaluating database of identifying an internationally authoritative speaker.

Description

technical field [0001] The invention relates to the field of voiceprint recognition, in particular to a speaker recognition method based on prior knowledge regularization covariance probabilistic linear discrimination analysis. Background technique [0002] Speaker recognition technology is a technology that uses the speaker's characteristic information contained in the speech signal to make a judgment and identify the true identity behind it. Speaker recognition technology has been widely used in identification, video conferencing, access control, military criminal investigation and many other fields, and has developed into an increasingly important modern biometric authentication technology. In recent years, the speaker recognition method based on the total variation factor has become the mainstream method in the field of speaker recognition, which does not strictly distinguish between speakers and channels, and models them as a whole. Through this technology, the first-o...

AI Technical Summary

Problems solved by technology

[0007] However, the limitation of the above algorithm framework is that the frame length and signal-to-noise ratio of each speech are different, and the probabilistic linear discriminant analysis model trained by using the global covariance matrix to describe the residual distribution will obviously be different from the The real model has a certain deviation, and will erase the useful information inherent in each sentence that can help improve recognition performance

Images