Neural Discriminant Analysis for Deep Speaker Embedding

Lantian Li, Dong Wang, Thomas Fang Zheng


Probabilistic Linear Discriminant Analysis (PLDA) is a popular tool in open-set classification/verification tasks. However, the Gaussian assumption underlying PLDA prevents it from being applied to situations where the data is clearly non-Gaussian. In this paper, we present a novel nonlinear version of PLDA named as Neural Discriminant Analysis (NDA). This model employs an invertible deep neural network to transform a complex distribution to a simple Gaussian, so that the linear Gaussian model can be readily established in the transformed space. We tested this NDA model on a speaker recognition task where the deep speaker vectors (x-vectors) are presumably non-Gaussian. Experimental results on two datasets demonstrate that NDA consistently outperforms PLDA, by handling the non-Gaussian distributions of the x-vectors.


 DOI: 10.21437/Interspeech.2020-2542

Cite as: Li, L., Wang, D., Zheng, T.F. (2020) Neural Discriminant Analysis for Deep Speaker Embedding. Proc. Interspeech 2020, 3251-3255, DOI: 10.21437/Interspeech.2020-2542.


@inproceedings{Li2020,
  author={Lantian Li and Dong Wang and Thomas Fang Zheng},
  title={{Neural Discriminant Analysis for Deep Speaker Embedding}},
  year=2020,
  booktitle={Proc. Interspeech 2020},
  pages={3251--3255},
  doi={10.21437/Interspeech.2020-2542},
  url={http://dx.doi.org/10.21437/Interspeech.2020-2542}
}