List your company

Spoken Instruction Following on Phone Noisy Audio (test)

65.4LC Win RateVIRBA
37.11244.45651.859.144Aug 25, 2025

Evaluation results

MethodLinks
VIRBABackbone=Qwen2.5-Omni2025.0865.4
VIRBABackbone=Qwen2-Audio2025.0861.9
Step-Audio-R12025.0858.6
Qwen2.5-Omni2025.0858.2
Single-view RL2025.0852.4
TTS-SFT2025.0849.7
Qwen2-Audio base2025.0838.2