List your company

BiLingual Speaker Recognition on TidyVoice

97.8Top-1 AccuracyCLIP ViT-B/32
59.00869.07979.1589.221Jun 2, 2026

Evaluation results

MethodLinks
CLIP ViT-B/32Backbone=ViT-B/322026.0697.8—
DinoV3 ViT-SBackbone=ViT-S2026.0696.8—
Expert features (MFCC) + linear SVMType=Expert Features2026.0693.8—
PlanFormerProtocol=Linear probing (5-fold CV), Parameters=1.99M, FLOPs=93.6MFLOPs, Classes=50, Type=Physical2026.0690.197.7
Random Init.Classes=50, Type=Physical2026.0660.5—