List your company

Automated Audio Captioning on AudioCaps (evaluation)

51.8SPIDErSLAM-AAC
25.38432.24239.145.958Oct 12, 2024Jan 25, 2025May 10, 2025Aug 24, 2025Dec 7, 2025Mar 22, 2026Jul 6, 2026

Evaluation results

MethodLinks
SLAM-AACPre-training Data=AC+CLp+WC+MA2024.1051.826.884.119.451.566.8
SLAM-LLMPre-training Data=AC+CL_P+WC+MA2026.0151.826.884.119.451.566.8
Tang et al.Pre-training Data=AC+CL+WC+LS+GS2024.1050.6—————
Tang et al.Pre-training Data=AC+CL+WC+LS+GS2026.0150.6—————
LOAEPre-training Data=AC+CL+WC2026.0150.526.781.619.350.466.2
EnCLAP-largePre-training Data=AC+CL2024.1049.525.580.318.849.965.5
EnCLAP-largePre-training Data=AC+CL2026.0149.525.580.318.849.965.5
WavCapsPre-training Data=AC+CL+WC2024.1048.52578.718.248.364.2
WavCapsPre-training Data=AC+CL+WC2026.0148.52578.718.248.364.2
SLAM–AACVariant=LoRA, Enc.=✓, LoRA=✓, Distill=–2026.0741.923.366.417.3——
SLAM–AACVariant=No LoRA, Enc.=✓, LoRA=✗, Distill=–2026.0738.723.759.817.5——
CARDVariant=CARD∗, Enc.=✗, LoRA=✓, Distill=Projector (early stages) + LLM2026.0735.221.255.415.1——
CARDVariant=Proj_Early, Enc.=✗, LoRA=✓, Distill=Early stages→Projector2026.0733.320.152.514.1——
CARDVariant=CARD⋄, Enc.=✗, LoRA=✓, Distill=Projector (all stages) + LLM2026.073220.349.914.1——
CARDVariant=No Distill, Enc.=✗, LoRA=✓, Distill=No teacher2026.0727.918.443.212.6——
CARDVariant=LLM Distill, Enc.=✗, LoRA=✓, Distill=LLM only2026.0727.818.443.512.2——
CARDVariant=Proj_Full, Enc.=✗, LoRA=✓, Distill=All stages→projector2026.0726.417.840.712.1——