# The IDLAB VoxSRC-20 Submission: Large Margin Fine-Tuning and Quality-Aware Score Calibration in DNN Based Speaker Verification

arXiv: https://arxiv.org/abs/2010.11255 · HTML: https://mlsanity.com/paper/2010.11255/

## Key facts

- Cited by 23 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 21 Oct 2020.

## Built on by (most cited first, 23 of 23)

- [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://mlsanity.com/paper/2110.13900/) (2021)
- [Autoregressive Diffusion Transformer for Text-to-Speech Synthesis](https://mlsanity.com/paper/2406.05551/) (2024)
- [Why does Self-Supervised Learning for Speech Recognition Benefit Speaker Recognition?](https://mlsanity.com/paper/2204.12765/) (2022)
- [LeBenchmark 2.0: a Standardized, Replicable and Enhanced Framework for Self-supervised Representations of French Speech](https://mlsanity.com/paper/2309.05472/) (2023)
- [The VoxCeleb Speaker Recognition Challenge: A Retrospective](https://mlsanity.com/paper/2408.14886/) (2024)
- [Learning Lip-Based Audio-Visual Speaker Embeddings with AV-HuBERT](https://mlsanity.com/paper/2205.07180/) (2022)
- [Leveraging ASR Pretrained Conformers for Speaker Verification through Transfer Learning and Knowledge Distillation](https://mlsanity.com/paper/2309.03019/) (2023)
- [Enhancing Speaker Verification with w2v-BERT 2.0 and Knowledge Distillation guided Structured Pruning](https://mlsanity.com/paper/2510.04213/) (2025)
- [Layer-aware TDNN: Speaker Recognition Using Multi-Layer Features from Pre-Trained Models](https://mlsanity.com/paper/2409.07770/) (2024)
- [DS-TDNN: Dual-stream Time-delay Neural Network with Global-aware Filter for Speaker Verification](https://mlsanity.com/paper/2303.11020/) (2023)
- [Probing the Information Encoded in Neural-based Acoustic Models of Automatic Speech Recognition Systems](https://mlsanity.com/paper/2402.19443/) (2024)
- [Rethinking Leveraging Pre-Trained Multi-Layer Representations for Speaker Verification](https://mlsanity.com/paper/2512.22148/) (2025)
- [Beyond Short Segments : Expanding Speaker Embeddings with Vector Archives](https://mlsanity.com/paper/2609.25007/) (2026)
- [The Royalflush System for VoxCeleb Speaker Recognition Challenge 2022](https://mlsanity.com/paper/2209.09010/) (2022)
- [THUEE system description for NIST 2020 SRE CTS challenge](https://mlsanity.com/paper/2210.06111/) (2022)
- [VoxWatch: An open-set speaker recognition benchmark on VoxCeleb](https://mlsanity.com/paper/2307.00169/) (2023)
- [ChinaTelecom System Description to VoxCeleb Speaker Recognition Challenge 2023](https://mlsanity.com/paper/2308.08181/) (2023)
- [Integrated Multi-Level Knowledge Distillation for Enhanced Speaker Verification](https://mlsanity.com/paper/2409.09389/) (2024)
- [Alpha Divergence Losses for Biometric Verification](https://mlsanity.com/paper/2511.13621/) (2025)
- [DAME: Duration-Aware Matryoshka Embedding for Duration-Robust Speaker Verification](https://mlsanity.com/paper/2601.13999/) (2026)
- [Kiwano: A Cutting-Edge Open-Source Toolkit for Speaker Verification](https://mlsanity.com/paper/2606.22369/) (2026)
- [Sparsity-Inducing Divergence Losses for Biometric Verification](https://mlsanity.com/paper/2606.31664/) (2026)
- [The Voiceprint Fallacy: Why Voices Are Not Unique Biometric Imprints](https://mlsanity.com/paper/2608.07980/) (2026)

## Abstract (arXiv)

In this paper we propose and analyse a large margin fine-tuning strategy and a quality-aware score calibration in text-independent speaker verification. Large margin fine-tuning is a secondary training stage for DNN based speaker verification systems trained with margin-based loss functions. It enables the network to create more robust speaker embeddings by enabling the use of longer training utterances in combination with a more aggressive margin penalty. Score calibration is a common practice in speaker verification systems to map output scores to well-calibrated log-likelihood-ratios, which can be converted to interpretable probabilities. By including quality features in the calibration system, the decision thresholds of the evaluation metrics become quality-dependent and more consistent across varying trial conditions. Applying both enhancements on the ECAPA-TDNN architecture leads to state-of-the-art results on all publicly available VoxCeleb1 test sets and contributed to our winning submissions in the supervised verification tracks of the VoxCeleb Speaker Recognition Challenge 2020.