# Real Additive Margin Softmax for Speaker Verification

arXiv: https://arxiv.org/abs/2110.09116 · HTML: https://mlsanity.com/paper/2110.09116/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 18 of its 25 references that are in the map.
- First posted on arXiv 18 Oct 2021.

## Builds on (most cited first, 18 of 18)

- [VoxCeleb2: Deep Speaker Recognition](https://mlsanity.com/paper/1806.05622/) (2018)
- [Generalized End-to-End Loss for Speaker Verification](https://mlsanity.com/paper/1710.10467/) (2017)
- [Additive Margin Softmax for Face Verification](https://mlsanity.com/paper/1801.05599/) (2018)
- [In defence of metric learning for speaker recognition](https://mlsanity.com/paper/2003.11982/) (2020)
- [Speaker Recognition Based on Deep Learning: An Overview](https://mlsanity.com/paper/2012.00931/) (2020)
- [Deep Speaker: an End-to-End Neural Speaker Embedding System](https://mlsanity.com/paper/1705.02304/) (2017)
- [BUT System Description to VoxCeleb Speaker Recognition Challenge 2019](https://mlsanity.com/paper/1910.12592/) (2019)
- [End-to-End Text-Dependent Speaker Verification](https://mlsanity.com/paper/1509.08062/) (2015)
- [Exploring the Encoding Layer and Loss Function in End-to-End Speaker and Language Recognition System](https://mlsanity.com/paper/1804.05160/) (2018)
- [Large Margin Softmax Loss for Speaker Verification](https://mlsanity.com/paper/1904.03479/) (2019)
- [Margin Matters: Towards More Discriminative Deep Neural Network Embeddings for Speaker Recognition](https://mlsanity.com/paper/1906.07317/) (2019)
- [CN-Celeb: multi-genre speaker recognition](https://mlsanity.com/paper/2012.12468/) (2020)
- [Centroid-based deep metric learning for speaker recognition](https://mlsanity.com/paper/1902.02375/) (2019)
- [TristouNet: Triplet Loss for Speaker Turn Embedding](https://mlsanity.com/paper/1609.04301/) (2016)
- [Unified Hypersphere Embedding for Speaker Recognition](https://mlsanity.com/paper/1807.08312/) (2018)
- [Deep Normalization for Speaker Vectors](https://mlsanity.com/paper/2004.04095/) (2020)
- [Max-margin Metric Learning for Speaker Recognition](https://mlsanity.com/paper/1510.05940/) (2015)
- [Gaussian-Constrained training for speaker verification](https://mlsanity.com/paper/1811.03258/) (2018)

## Built on by (most cited first, 3 of 3)

- [Hyperbolic Additive Margin Softmax with Hierarchical Information for Speaker Verification](https://mlsanity.com/paper/2601.19709/) (2026)
- [Memory-Efficient Training for Deep Speaker Embedding Learning in Speaker Verification](https://mlsanity.com/paper/2412.01195/) (2024)
- [Spoken Language Identification with Pre-trained Models and Margin Loss](https://mlsanity.com/paper/2605.01905/) (2026)

## Abstract (arXiv)

The additive margin softmax (AM-Softmax) loss has delivered remarkable performance in speaker verification. A supposed behavior of AM-Softmax is that it can shrink within-class variation by putting emphasis on target logits, which in turn improves margin between target and non-target classes. In this paper, we conduct a careful analysis on the behavior of AM-Softmax loss, and show that this loss does not implement real max-margin training. Based on this observation, we present a Real AM-Softmax loss which involves a true margin function in the softmax training. Experiments conducted on VoxCeleb1, SITW and CNCeleb demonstrated that the corrected AM-Softmax loss consistently outperforms the original one. The code has been released at https://gitlab.com/csltstu/sunine.