ReDimNet2+: Multi-Corpus Data Scaling for Robust Speaker Verification

2026-09-29

AI Summary

This study addresses the limited robustness of speaker verification in complex scenarios involving cross-device and cross-room conditions, as well as compressed transmission. To this end, we propose a multi-strategy collaborative optimization framework. Methodologically, ReDimNet2 serves as the backbone network, combined with multi-corpus expansion and codec- and waveform-level augmentation to enhance data diversity. Furthermore, spectral coloring analysis is introduced to guide large-margin fine-tuning (LMFT), and a graph retrieval-based re-ranking strategy is designed to refine decision boundaries. Experimental results demonstrate that the proposed system reduces the equal error rate (EER) to 0.82% on the VoxCeleb1 test set and achieves an EER of only 1.99% under stress testing. These results significantly outperform existing baselines, effectively improving verification reliability in open-world scenarios.

Abstract

Automatic speaker verification must remain reliable across devices, rooms, and compression pipelines. We present ReDimNet2+, which scales training of the compact ReDimNet2 backbone across seven public corpora (63,934 speakers, about 8,675 hours). Analysis of a VoxBlink2 subset reveals a shift in predicted spectral coloration, motivating codec and waveform augmentation alongside this multi-corpus training, large-margin fine-tuning (LMFT), and graph-based retrieval reranking. With random 4-second evaluation windows for all models, ReDimNet2+ LMFT reduces pooled VoxCeleb1 EER from 2.42% to 0.82% and a 26-condition robustness stress-test EER from 7.21% to 1.99%. Under this shared local protocol, it reaches 0.35% EER on VoxCeleb1-O versus 0.787% for the best evaluated WeSpeaker checkpoint. On a VoxBlink2 retrieval subset, reranking improves the final model's Pr@k from 0.7413 to 0.7687.

Problem

• Speaker Verification
• Robustness
• Multi-Corpus
• Spectral Coloration
• Equal Error Rate

Innovation

• Multi-Corpus Data Scaling
• Codec and Waveform Augmentation
• Large-Margin Fine-Tuning
• Graph-based Retrieval Reranking
• Robust Speaker Verification