The project's main aim is to make archival sound recordings at the Uppsala University Library more accessible and useful to researchers and ensure that they a) keep their functionality as a resource for the future and b) are preserved and accessed according to jurisdictional and ethical regulations. This aim could be reached by improving and making more efficient the description and classification of archival sound recordings at the library. Material: The recordings are part of the collections of digital personal papers in the Special Collections of Uppsala University Library. They consists of interviews and speeches from around the years 1950 - 2000. They have emerged from various contexts and by different creators and is only partly described on a very superficial level. The problem is that much of the content of the recordings is still unknown, at least at a detailed level, and sensitive information about living persons could exists without the library staff having knowledge about it. A thorough knowledge of the content of the recordings is thus needed by the library to correctly be able to describe the data and judge the sensitivity of the recordings. However, it takes too much time to process this manually, which is why automatic transcription with Whisper is desired. Method: Transcribe recorded speech with Whisper. Transcriptions could then be used in various ways by library staff and researchers to describe and analyze the material.