From close-listening to distant-listening: Developing tools for large-scale analysis of (Danish) radio and audio media archives

This project is about developing tools for analyzing audio media content and in particular Danish music radio. It is based on the radio collection of The Royal Danish Library and is guided by the following two research questions:

RQ1: How has the distribution of music and talk on DR’s radio channel P3 developed 1989-2019?

  • How much music and how much talk? (classification)
  • How long are the sequences with talk/music? 
  • Development of amount of silence/pauses in the data?

RQ2: Distribution of gender in the talk (and maybe also the music) on P3 1989-2019. 

  • How much time with male voices and female voices respectively?
  • Development of tempo, pitch, volume and other voice-expression related features?

Background: The pilot project is building on previous research of the development of the everyday Danish morning music radio show Go’ Morgen P3from 1989 to 2016. This project was a part of the large collective project A Century of Radio and Music in Denmarkfinanced by the Independent Research Fund Denmark (2013-2018).

Method:The pilot project is zooming out from five case studies in the Go’ Morgen P3study to the entire P3 programming 1989-2019. And instead of using human ears to listen closely to single programs, the ‘listening’ (or signal processing) is done by a super-computer. Methodologically this is a shift from close-listeningto a few programs to large-scaledistant-listeningto more than 200.000 hours of radio.

Context:The analytical results from the pilot study will be contextualized by a study of DR’s (the national Danish radio) political and cultural conditions and development 1989-2019.

Scope: The pilot project focuses on radio but the tools developed might also be used for analyzing and classifying audio content in other kinds of audio media like audiobooks, soundtracks for audiovisual media and podcasts.

The project is led by Iben Have