VAE-based regularization for deep speaker embedding
Sound
2019-04-09 v1 Machine Learning
Audio and Speech Processing
Abstract
Deep speaker embedding has achieved state-of-the-art performance in speaker recognition. A potential problem of these embedded vectors (called `x-vectors') are not Gaussian, causing performance degradation with the famous PLDA back-end scoring. In this paper, we propose a regularization approach based on Variational Auto-Encoder (VAE). This model transforms x-vectors to a latent space where mapped latent codes are more Gaussian, hence more suitable for PLDA scoring.
Cite
@article{arxiv.1904.03617,
title = {VAE-based regularization for deep speaker embedding},
author = {Yang Zhang and Lantian Li and Dong Wang},
journal= {arXiv preprint arXiv:1904.03617},
year = {2019}
}