AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Disentangling Voice and Content with Self-Supervision for Speaker Recognition

Tianchi Liu, Kong Aik Lee, Qiongqiong Wang, Haizhou Li

NeurIPS 2023 · Label-efficient and robust learning › Recognition and long-tail · approach: Multilingual language representation pretraining · 4 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace TSP with Xi aggregationtResNet34 · VoxCeleb1/SITWEER and minDCF: 5.02%/4.19% average reductionsImproved
replace Xi with RecXitResNet34 · VoxCeleb1/SITWEER and minDCF: 6.96%/5.82% average reductionsImproved
replace Xi with RecXiECAPA-TDNN · VoxCeleb1/SITWEER and minDCF: 12.15%/10.66%Improved
replace Chan.&Con. with XiECAPA-TDNN · VoxCeleb1-OEER: 1.377 -> 1.340Mixed
replace ResNet34 with tResNet34TSP aggregation · VoxCeleb1-HEER: 2.500 -> 2.257Improved
add self-supervised speaker-preserving loss L_ssp (ablation)RecXi · VoxCeleb1/SITWEER and minDCF: 5.07%/5.44% average reductionsImproved
add L_ssp (ablation)RecXi(phi, phi_lin) ECAPA · SITW evalEER: 2.351 -> 2.105Improved
use phi only instead of phi and phi_lin, without L_ssp (ablation)RecXi tResNet · VoxCeleb1-HEER: 2.130 -> 2.185Hurt
use phi only instead of phi and phi_lin, with L_ssp (ablation)RecXi tResNet · VoxCeleb1-HEER: 2.097 -> 2.117No clear effect
use MSE loss instead of similarity-preserving loss for L_ssp (ablation)RecXi ECAPA · SITW evalEER: 2.105 -> 2.292Hurt
MSE L_ssp vs no L_ssp (ablation)RecXi ECAPA · SITW evalEER: 2.351 -> 2.292Improved
reduce transition matrices from 16 to 1 (ablation)RecXi · SITW evalEER: 1.832 -> 3.882Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy