FIAT/IFTA World Conference 2025

Since the advent of the Internet mid 1990’s, free text search has been a central tool. First via indexes of online webpages in readable formats, then all sorts of digitized materials made searchable via OCR. However, written sources are by definition secondary sources, not least in relation to media archives. In this paper, we will demonstrate how the use of free sound search has been instrumental in two recent research projects on Danish music history, conducted at the Royal Danish Library.

The base is the tool xcorrSound. Via indexing of sound archives, it is possible to match sound files and perform searches into big amounts of data. Via a custom-built interface, the results are delivered in tables with relevant data, including a direct link to the file in the media archive and a custom-built media player, directly executing the files.

In this case, we used it to identify the use of specific songs in Danish radio and television from 1989 to 2020, indexed across the Danish media collections. From the data we could not only map airplay of specific songs year by year within minutes, but also analyze the contexts to a much higher degree than before. For instance, how a specific track was introduced or how the reception of a given song changes over time.

The tool has changed generally dark archives into vivid and rich resources for re-telling Danish media history. For now, the searches are performed track by track, but the perspectives of AI implementations are evident.