Recently, there has been a growing trend of adopting 3D digital humans (also referred to as virtual humans) in the production of broadcast and video content. Due to their controllability and ability to be utilized without temporal or spatial constraints, digital humans (or virtual humans) are emerging as an effective solution for enhancing production efficiency. In particular, the production of digital humans is shifting from traditional 2D formats to interactive 3D forms, and accordingly, various datasets for creating 3D digital humans are now being made available on public data platforms such as AI Hub. However, there is currently a lack of facial data necessary for generating speech-driven 3D facial animation videos that can realistically express Korean speech through 3D digital humans (or virtual humans). This study aims to address this gap by constructing and releasing a dataset specifically designed for producing 3D digital humans (or virtual humans) capable of natural Korean speech. By doing so, we seek to reduce production costs through the application of generative AI and contribute to the advancement of related technologies and the activation of the associated industry ecosystem. Furthermore, this research aligns with the national AI policy initiative of "AI in Everyday Life," corresponds with recent R&D priorities combining AI and the metaverse, and is expected to be applicable to the development of the Digital Platform Government in the future.