Hi, KIA
收藏资源简介:
Hi, KIA是一个专为唤醒词(WUW)设计的语音情感识别数据集,由韩国科学技术院创建。该数据集包含488条来自八位韩国演员的情感标注语音,每条语音标注有愤怒、快乐、悲伤或中性四种情感状态。数据集的创建过程包括场景选择、录音/后处理和人工验证标签一致性。Hi, KIA数据集特别适用于车载语音用户界面(VUIs),也可用于一般目的的文本依赖语音情感识别,旨在提升人机交互的情感理解。
Hi, KIA is a speech emotion recognition dataset specifically designed for Wake-Up Word (WUW) tasks, developed by the Korea Advanced Institute of Science and Technology (KAIST). This dataset includes 488 emotionally annotated speech utterances from eight Korean actors, with each utterance labeled with one of four emotional states: anger, happiness, sadness, or neutrality. The dataset construction process encompasses scene selection, audio recording and post-processing, as well as manual verification of label consistency. The Hi, KIA dataset is particularly applicable to in-vehicle Voice User Interfaces (VUIs), and can also be utilized for general-purpose text-dependent speech emotion recognition, aiming to enhance emotional understanding in human-computer interaction.




