UTeMo audiovisual dataset for emotion recognition

UTeMo is a database recorded with statements lexically vocalized in a Mexican variant of the Spanish language, which is specifically designed for multi-modal (audiovisual) emotion recognition (using vocal and facial expressions). It comprises 1801 video samples with a total of 105 minutes. It is composed of high quality data, as it supplies high resolution images (1920x1080 pixels at 30 fps) and high fidelity audio (sampling rate of 48KHz) files. UTeMo can be considered as a database whose number of samples is approximately balanced according to the seven emotion classes (sadness, surprise, joy, anger, fear, disgust and neutral), so every emotional state is well represented.UTeMo是首个采用墨西哥西班牙语语音表述录制语句的数据库，专门面向基于语音与面部表情的多模态（multi-modal）音视频（audiovisual）情感识别任务。该数据集包含1801个视频样本，总时长共计105分钟。其数据品质优异，提供了分辨率为1920×1080像素、帧率30fps的高分辨率图像，以及采样率为48kHz的高保真音频文件。 UTeMo可被视为样本分布均衡的数据库，涵盖悲伤、惊讶、喜悦、愤怒、恐惧、厌恶与中性共七大情感类别，每一类情感状态均得到了充分的样本覆盖。