# Deep Normalization for Speaker Vectors

arXiv: https://arxiv.org/abs/2004.04095 · HTML: https://mlsanity.com/paper/2004.04095/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 7 Apr 2020.

## Built on by (most cited first, 4 of 4)

- [Real Additive Margin Softmax for Speaker Verification](https://mlsanity.com/paper/2110.09116/) (2021)
- [TMS: A Temporal Multi-scale Backbone Design for Speaker Embedding](https://mlsanity.com/paper/2203.09098/) (2022)
- [Label-free Knowledge Distillation with Contrastive Loss for Light-weight Speaker Recognition](https://mlsanity.com/paper/2212.03090/) (2022)
- [Interpolating Speaker Identities in Embedding Space for Data Expansion](https://mlsanity.com/paper/2508.19210/) (2025)

## Abstract (arXiv)

Deep speaker embedding has demonstrated state-of-the-art performance in speaker recognition tasks. However, one potential issue with this approach is that the speaker vectors derived from deep embedding models tend to be non-Gaussian for each individual speaker, and non-homogeneous for distributions of different speakers. These irregular distributions can seriously impact speaker recognition performance, especially with the popular PLDA scoring method, which assumes homogeneous Gaussian distribution. In this paper, we argue that deep speaker vectors require deep normalization, and propose a deep normalization approach based on a novel discriminative normalization flow (DNF) model. We demonstrate the effectiveness of the proposed approach with experiments using the widely used SITW and CNCeleb corpora. In these experiments, the DNF-based normalization delivered substantial performance gains and also showed strong generalization capability in out-of-domain tests.