This release is for the full corpus (i.e., all 4,834 recordings). In addition to the recordings, the release contains spectrograms of each recording and a spreadsheet (HuMS.csv) with the acoustic, linguistic, and perceptual analyses of each recording. The perceptual ratings come from a set (n = 5) of internal raters, who rated all stimuli in the corpus.
The HuMS-A1 contains 160 recordings - the top 20 and bottom 20 recordings in terms of perceived musicality for the categories of talker sex (male, female) and intended audience (child-directed, adult-directed). The download includes the 160 recordings, as well as a data file (HuMS-A1.csv) containing the acoustic, linguistic, and perceptual ratings from both the smaller set of trusted raters (n = 5) as well as a larger group of raters (total n = 100; n = 50 per talker sex).
The HuMS-A2 contains 420 recordings - the top 15 and bottom 15 recordings in terms of perceived musicality for each unqiue talker in the corpus (n = 14). The download includes the 420 recordings, as well as a data file (HuMS-A2.csv) containing the acoustic, linguistic, and perceptual ratings from both the smaller set of trusted raters (n = 5) as well as a larger group of raters (total n = 320; n = 18 to 27 per talker).
