Speaker
Benjamin Beuster
(Sikt - Norwegian agency for shared services in education and research)
Description
Social research increasingly includes media formats like audio and video, which are often poorly documented and inaccessible. While archives handle traditional survey data well, media files are mostly limited to minimally annotated zip files due to the complexity of proper documentation. Recent advancements in AI, including the Whisper model, along with the use of Pydantic models and structured output, now allow for rapid metadata extraction, transcription, and reliable, structured summaries.
This presentation demonstrates how AI can streamline the documentation and conversion of media files into the DDI-Lifecycle "Other Material" format, significantly improving accessibility and usability for researchers and data archives.
Primary author
Benjamin Beuster
(Sikt - Norwegian agency for shared services in education and research)