Table 4.

Comparative experiment details.

Model	Speech	Text	Vision	Fusion
MFN [43]	LSTM	LSTM	LSTM	Feature fusion
MCTN [32]	CNN	CNN	CNN	Concatenation
RAVEN [33]	LSTM	LSTM	LSTM	Feature fusion
HFusion [34]	openSMILE	Word2vec + CNN	3D CNN	Hierarchical fusion
MulT [35]	Conv 1D	Conv 1D	Conv 1D	Feature fusion
SSE-FT [40]	Wav2vec	Roberta	FabNet	Hierarchical fusion
CMC-HF(ours)	CSA encoder	EFC encoder	PC encoder	Hierarchical fusion