search

ddlBoJack / emotion2vec

[ACL 2024] Official PyTorch code for extracting features and training downstream models with emotion2vec: Self-Supervised Pre-Training for Speech Emotion Representation
649 stars 47 forks source link

setup for 4-primary classes #50

Open KeiKinn opened 2 weeks ago

KeiKinn commented 2 weeks ago

hi, could you share the setup for the 4-primary classification shown in the performance radar?

The upstream is emotion2vec_plus_large, but how about the downstream? Do you use an universal classifier? or?

Thank you

ddlBoJack commented 1 week ago

Just inference with emotion2vec_plus_large, and compute on the 4-primary class as these emotions are shared by different datasets.

KeiKinn commented 1 week ago

hi, thank you for your reply. emotion2vec_plus_large has two versions: 5-class and 9-class. And I would persume that you used the first one (four emotion and unknown). Did you exclude the 'unkonwn' case when get the metrics?