Speaker recognition method based on dictionary learning and low rank matrix decomposition

Abstract

The invention provides a speaker recognition method based on dictionary learning and low rank matrix decomposition. Thespeaker recognition method comprises the following steps that step 1, speaker audiosare subjected to processing of pre-emphasis, framing, windowing, endpoint detection and the like; step 2, MFCC features corresponding tosentences of speakersare extracted, and a GMM-UBM model is trained; step 3, a global difference spatial matrix T and a global difference spatial factor w are estimated by joint factor analysis (JFA); step 4,i-vector of the sentences of each speaker is obtained; and step 5, the i-vector of the M dimension isextracted from a training set to form a feature matrix, and adistinguish dictionary isgenerated according to the training set and a test set, the obtained dictionary is taken as abackend processing and scoring moduleof thei-vector, anda basis isprovided for the final distinguish. Coding coefficients adapted to the dictionary learning criterion can effectively improve the recognition power, and optimal classification is carried out through structured sparseness.

Description

technical field [0001] The invention relates to the field of speaker recognition, in particular to a back-end i-vector dictionary discrimination method for a speaker recognition system. Background technique [0002] Speaker Recognition (SR), also known as voiceprint recognition, is a biometric authentication technology that uses specific speaker information contained in voice signals to identify the identity of the speaker. In recent years, the introduction of the identity vector (i-vector) speaker modeling method based on factor analysis has significantly improved the performance of the speaker recognition system. Experiments show that in the factor analysis of the speaker's speech, usually the channel subspace will contain the speaker's information. Therefore, i-vector uses a low-dimensional total variable space to represent the speaker subspace and channel subspace, and maps the speaker's voice to this space to obtain a fixed-length vector representation (i.e., i-vector)...

AI Technical Summary

Problems solved by technology

Under unconstrained conditions, it is difficult to guarantee the stability of the algorithm, especially when the noise pollution is too large, which will reduce the recognizability of the trained dictionary

Images