Akan Cinematic Emotions (ACE)
收藏资源简介:
Akan Cinematic Emotions (ACE)数据集是由哥本哈根信息技术大学、哥伦比亚大学和苹果机器学习研究机构共同开发的,包含385个情感标记的对话和6162条发言,涵盖音频、视觉和文本多模态信息。该数据集以非洲语言Akan为对象,致力于填补情感识别研究中低资源语言的空白。数据集通过21部Akan语电影对话进行收集和标注,包含情感标签和单词级别的韵律突出标注,适用于低资源语言和跨文化的情感识别研究。
The Akan Cinematic Emotions (ACE) dataset was co-developed by the IT University of Copenhagen, Columbia University, and Apple Machine Learning Research. It comprises 385 emotion-annotated dialogues and 6,162 utterances, encompassing audio, visual, and textual multimodal data. Focused on the African language Akan, this dataset aims to fill the critical gap in emotion recognition research for low-resource languages. Collected and annotated from dialogues in 21 Akan-language films, the dataset includes both emotion labels and word-level prosodic prominence annotations, making it suitable for emotion recognition research on low-resource languages and cross-cultural studies.




