The IDLAB Voxsrc-20 Submission: Large Margin Fine-tuning And Quality-aware Score Calibration In DNN Based Speaker Verification
2020 · Jenthe Thienpondt, Brecht Desplanques, Kris Demuynck
Abstract
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 winn
Authors
(none)
Tags
Stats
Related papers
- Tackling The Score Shift In Cross-lingual Speaker Verification By Exploiting Language Information (2021)3.58
- Investigation Of Different Calibration Methods For Deep Speaker Embedding Based Verification Systems (2022)0.00
- Margin Matters: Towards More Discriminative Deep Neural Network Embeddings For Speaker Recognition (2019)15.25
- Large Margin Softmax Loss For Speaker Verification (2019)14.66
- Neural Scoring: A Refreshed End-to-end Approach For Speaker Recognition In Complex Conditions (2024)0.00
- An Attention-based Backend Allowing Efficient Fine-tuning Of Transformer Models For Speaker Verification (2022)11.08
- Experimenting With Additive Margins For Contrastive Self-supervised Speaker Verification (2023)4.52
- Delving Into Voxceleb: Environment Invariant Speaker Recognition (2019)10.35