Mandarin Chinese Emotional Speech Dataset - Portrayed (MES-P)
收藏资源简介:
MES-P是一个专门为研究普通话情感语音而设计的数据集,由苏州大学光电科学与工程学院的研究人员创建。该数据集包含5376条情感语音样本,涵盖了喜悦、愤怒和悲伤三种情感的温和与强烈版本,以及中性语音。MES-P的独特之处在于它不仅记录了听众感知的情感(近端标签),还记录了说话者意图表达的情感(远端标签),从而使得研究人类情感智能成为可能。数据集的创建过程中,说话者被要求根据预定义的文本脚本朗读,以确保语音样本的情感表达与日常普通话的音调分布相匹配。此外,MES-P还捕捉了情感强度的变化,为情感识别和理解提供了丰富的资源。该数据集的应用领域包括人机交互、多媒体索引和情感监控等。
MES-P is a dataset specifically designed for research on Mandarin emotional speech, created by researchers from the School of Optoelectronic Science and Engineering of Soochow University. This dataset contains 5376 emotional speech samples, covering mild and intense versions of three emotions: joy, anger, and sadness, as well as neutral speech. What makes MES-P unique is that it not only records the emotions perceived by listeners (proximal labels) but also the emotions that speakers intend to express (distal labels), thus enabling research on human emotional intelligence. During the dataset creation process, speakers were asked to read aloud based on predefined text scripts to ensure that the emotional expressions of the speech samples match the pitch distribution of daily Mandarin. In addition, MES-P also captures changes in emotional intensity, providing rich resources for emotion recognition and understanding. The application fields of this dataset include human-computer interaction, multimedia indexing, emotional monitoring, and so on.

- 1MES-P: an Emotional Tonal Speech Dataset in Mandarin Chinese with Distal and Proximal Labels苏州大学光电科学与工程学院 · 2018年



