The Idlab Voxsrc-20 Submission: Large Margin Fine-Tuning and Quality-Aware Score Calibration in DNN Based Speaker Verification

By Jenthe Thienpondt, Brecht Desplanques, Kris Demuynck. 2020

115 citations recorded by the source corpus.

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 Vox-Celeb Speaker Recognition Challenge 2020.

Subfields: Artificial Intelligence, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Jinyu Li
2. Shujie Liu
3. Zhuo Chen
4. Yu Wu
5. Takuya Yoshioka
6. Sanyuan Chen
7. Furu Wei
8. Jian Wu

In Signal Processing

1. Jinyu Li
2. Zhuo Chen
3. Takuya Yoshioka
4. Shujie Liu
5. Yu Wu
6. Sanyuan Chen
7. Naoyuki Kanda
8. Jian Wu

Citations in the corpus

• WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing (2021, methodology): “To further push the limit of the speaker verification system, we introduce the large margin fine-tuning and quality-aware score calibration strategies [58] into our best systems and the corresponding results are listed at the bottom of Table II.”
• The SpeakIn System for VoxCeleb Speaker Recognition Challange 2021 (2021, methodology): “After doing AS-Norm and QMF, our system’s EER improved 19.6% relatively, and minDCF improved 21.76% relatively compared to the finetuned system.”
• Wespeaker: A Research and Production Oriented Speaker Embedding Learning Toolkit (2022, background): “The large margin fine-tuning strategy was first proposed in [26] and widely used in speaker verification challenge systems [19, 27, 28] to further enhance the system’s performance.”