Lip-Motion/Voice Recognition Videos

This consists of 20 randomly selected individuals who were videotaped reciting the FitnessGram Pacer Test script and an arbitrary sentence in two separate videos respectively. Each person has two videos in each of their folder, which has been labeled with numbers to preserve the privacy of their names. Originally, this data was collected to train/test a compound biometric system that consists of lip-motion and voice authentication. The training videos averaged between 30 to 45 seconds and the testing videos averaged between 5 to 8 seconds each.

本数据集包含20名随机遴选的受试者，每位受试者均录制了两段独立视频：一段为诵读FitnessGram Pacer Test脚本的视频，另一段为诵读任意语句的视频。每位受试者的文件夹内均存储有两段视频，文件夹以数字编号以保护其姓名隐私。该数据集最初采集的目的为训练与测试一套融合唇动识别与语音认证的复合生物识别系统。训练用视频的平均时长介于30至45秒之间，测试用视频的平均时长则介于5至8秒之间。