You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.
tal bf58aefe54 add ctal 1 year ago
..
calculate_eer.py add ctal 1 year ago
m2p_dataloader.py add ctal 1 year ago
m2p_finetune.py add ctal 1 year ago
m2p_mask.py add ctal 1 year ago
m2p_model.py add ctal 1 year ago
m2p_optimization.py add ctal 1 year ago
m2p_runner.py add ctal 1 year ago
m2p_transfer.py add ctal 1 year ago
run_m2pretrain.py add ctal 1 year ago
run_s-iemocap.py add ctal 1 year ago
run_s-pretrain.py add ctal 1 year ago
run_s-reg-mosei.py add ctal 1 year ago

该算法提出一个新的基于音频和文本的跨模态预训练模型, CTAL。通过大量的音频和文本对的两个代理任务来学习音频和文本之间的模态内和模态间的联系:屏蔽的语言建模和屏蔽的跨模态声学建模。在对多个下游音频和文本任务进行微调后,CTAL 模型在不同任务上有明显的改进,包括情感分类、情绪分析和说话人验证。其中在 IEMOCAP(Emotion Classification) 数据集上WA 达到73.95%,在MOSEI(Sentiment Analysis)上 F1 达到 81.01%。

Text Python

Contributors (2)