Cross-cultural music corpus: The Natural History of Song Discography (randomized 14s excerpts)

This repository contains 14-second excerpts of the Natural History of Song Discography. The audio corpus was originally published in Mehr et al. (2019, Science) and the full audio files are available via DUA on the Harvard Dataverse (https://doi.org/10.7910/DVN/SESAO1). This repository makes available the excerpts of each track, under Fair Use, given their brief length (14 seconds each). These were also used in other papers (e.g., Mehr & Singh et al., 2018, Current Biology; Bainbridge & Bertolo et al., 2021, Nature Human Behaviour; Hilton & Crowley-de Thierry et al., 2022, JEP:General; Yurdum et al., 2023, PNAS) and are concurrently posted on the Open Science Framework (https://osf.io/vcybz/). They are permanently archived here for easier accessibility.

Other Natural History of Song data, including staff-notation transcriptions of all the songs, are available at https://osf.io/jmv3q/. There is also an expanded version of the corpus, with 1007 songs (including all 118 of the original NHS songs, available on Zenodo at https://doi.org/10.5281/zenodo.8237500.

The audio excerpts are provided in MP3 format for use in perception experiments. There are 118 excerpts. Each row of NHSDiscography-metadata.csv corresponds to one song, and provides general information about each song, including original sources. NHSDiscography-codebook.pdf summarizes the types of metadata available. Both these documents were originally published in Mehr et al. (2019, Science), along with the original corpus, and quite a bit of other information concerning how the corpus was built (https://www.science.org/doi/10.1126/science.aax0868). 

Please contact Samuel Mehr (sam@yale.edu) if you have questions concerning the use of these audio files.