List your company

Text-Audio Alignment on Baton-Pair three-sound-event

69.1Pair AccuracyAQAScore
20.7433.29545.8558.405Jan 21, 2026

Evaluation results

MethodLinks
AQAScoreBackbone=Qwen2.5-Omni-3B2026.0169.1
AQAScoreBackbone=Qwen2.5-Omni-7B2026.0167.8
AQAScoreBackbone=AF32026.0163.2
AQAScoreBackbone=AF3-Think2026.0162
CLAPScoreVariant=630k-best2026.0161.9
AQAScoreBackbone=AF3-Chat2026.0160.2
CLAPScoreTraining Data=Music+AS2026.0160
CLAPScoreTraining Data=Music+Speech+AS2026.0158
CLAPScoreTraining Data=630k-AS2026.0157.5
CLAPScoreTraining Data=Music+Speech2026.0156.8
PromptingBackbone=Qwen2.5-Omni-7B2026.0154.3
PromptingBackbone=AF32026.0150.7
PromptingBackbone=AF3-Chat2026.0150.5
PromptingBackbone=Qwen2.5-Omni-3B2026.0150.1
PromptingBackbone=AF3-Think2026.0150
CascadeBackbone=AF3-Chat2026.0142
CascadeBackbone=Qwen2.5-Omni-7B2026.0138.2
CascadeBackbone=AF32026.0138
CascadeBackbone=Qwen2.5-Omni-3B2026.0135.6
CascadeBackbone=AF3-Think2026.0122.6