SpeakingFaces

SpeakingFaces是由智能系统与人工智能研究所创建的大型多模态数据集，包含142名参与者的同步热成像、视觉和音频数据，总计约3.8TB。数据集通过高分辨率的热成像和视觉图像流与每个参与者的音频记录同步，用于支持机器学习研究，特别是在人机交互、生物识别认证和识别系统等领域。数据集的创建过程涉及从多个角度记录参与者，并使用公开可用的命令集进行语音指令的录制。SpeakingFaces数据集的应用领域广泛，旨在解决多模态数据处理和分析中的挑战，特别是在非理想物理环境中的人机交互和语音识别问题。

SpeakingFaces is a large-scale multimodal dataset created by the Institute of Intelligent Systems and Artificial Intelligence. It contains synchronized thermal imaging, visual and audio data from 142 participants, with a total size of approximately 3.8 TB. The dataset synchronizes high-resolution thermal imaging and visual image streams with audio recordings of each participant, and is used to support machine learning research, particularly in fields such as human-computer interaction, biometric authentication and recognition systems. The creation of the dataset involves recording participants from multiple angles and using publicly available command sets to record speech instructions. The SpeakingFaces dataset has a wide range of application areas, aiming to address challenges in multimodal data processing and analysis, particularly human-computer interaction and speech recognition in non-ideal physical environments.