# Attention Back-end for Automatic Speaker Verification with Multiple Enrollment Utterances

arXiv: https://arxiv.org/abs/2104.01541 · HTML: https://mlsanity.com/paper/2104.01541/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 4 Apr 2021.

## Built on by (most cited first, 8 of 8)

- [SafeEar: Content Privacy-Preserving Audio Deepfake Detection](https://mlsanity.com/paper/2409.09272/) (2024)
- [Disentangling Voice and Content with Self-Supervision for Speaker Recognition](https://mlsanity.com/paper/2310.01128/) (2023)
- [SVSNet: An End-to-end Speaker Voice Similarity Assessment Model](https://mlsanity.com/paper/2107.09392/) (2021)
- [The First VoicePrivacy Attacker Challenge](https://mlsanity.com/paper/2504.14183/) (2025)
- [Probabilistic Back-ends for Online Speaker Recognition and Clustering](https://mlsanity.com/paper/2302.09523/) (2023)
- [The First VoicePrivacy Attacker Challenge Evaluation Plan](https://mlsanity.com/paper/2410.07428/) (2024)
- [Cluster-Guided Unsupervised Domain Adaptation for Deep Speaker Embedding](https://mlsanity.com/paper/2303.15944/) (2023)
- [EchoFake: A Replay-Aware Dataset for Practical Speech Deepfake Detection](https://mlsanity.com/paper/2510.19414/) (2025)

## Abstract (arXiv)

Probabilistic linear discriminant analysis (PLDA) or cosine similarity have been widely used in traditional speaker verification systems as back-end techniques to measure pairwise similarities. To make better use of multiple enrollment utterances, we propose a novel attention back-end model, which can be used for both text-independent (TI) and text-dependent (TD) speaker verification, and employ scaled-dot self-attention and feed-forward self-attention networks as architectures that learn the intra-relationships of the enrollment utterances. In order to verify the proposed attention back-end, we conduct a series of experiments on CNCeleb and VoxCeleb datasets by combining it with several sate-of-the-art speaker encoders including TDNN and ResNet. Experimental results using multiple enrollment utterances on CNCeleb show that the proposed attention back-end model leads to lower EER and minDCF score than the PLDA and cosine similarity counterparts for each speaker encoder and an experiment on VoxCeleb indicate that our model can be used even for single enrollment case.