Author(s)

Wei Cao, Chunyan Liang^*, Shuxin Cao

College of Computer Science and Technology, Shandong University of Technology, Zibo, China.

Due to differences in the distribution of scores for different trials, the performance of a speaker verification system will be seriously diminished if raw scores are directly used for detection with a unified threshold value. As such, the scores must be normalized. To tackle the shortcomings of score normalization methods, we propose a speaker verification system based on log-likelihood normalization (LLN). Without a priori knowledge, LLN increases the separation between scores of target and non-target speaker models, so as to improve score aliasing of “same-speaker” and “different-speaker” trials corresponding to the same test speech, enabling better discrimination and decision capability. The experiment shows that LLN is an effective method of scoring normalization.

Speaker Verification, Score Normalization, Log-Likelihood Normalization, Zero Normalization, Test Normalization

Cao, W. , Liang, C. and Cao, S. (2020) Speaker Verification Based on Log-Likelihood Score Normalization. Journal of Computer and Communications, 8, 80-87. doi: 10.4236/jcc.2020.811006.