A Generative Model for Score Normalization in Speaker Recognition
This addresses the issue of data-set shift in speaker recognition systems, though it is incremental as it builds on existing normalization methods.
The paper tackles the problem of score normalization in speaker recognition by proposing a generative model based on probability theory, achieving improvements comparable to ZT-norm on the text-dependent RSR 2015 database.
We propose a theoretical framework for thinking about score normalization, which confirms that normalization is not needed under (admittedly fragile) ideal conditions. If, however, these conditions are not met, e.g. under data-set shift between training and runtime, our theory reveals dependencies between scores that could be exploited by strategies such as score normalization. Indeed, it has been demonstrated over and over experimentally, that various ad-hoc score normalization recipes do work. We present a first attempt at using probability theory to design a generative score-space normalization model which gives similar improvements to ZT-norm on the text-dependent RSR 2015 database.