List your company

Spoken Instruction Following on Accent OOD (test)

0.689Win Rate (LC)VIRBA
0.404040.478020.5520.62598Aug 25, 2025

Evaluation results

MethodLinks
VIRBABackbone=Qwen2.5-Omni2025.080.689
VIRBABackbone=Qwen2-Audio2025.080.657
Step-Audio-R12025.080.627
Qwen2.5-Omni2025.080.61
Single-view RL2025.080.565
TTS-SFT2025.080.528
Qwen2-Audio base2025.080.415