Headphones–Speakers: A multimodal dataset of human standstill and micromotion to music

A multimodal dataset of spontaneous body micromotion while listening to music over headphones versus loudspeakers (University of Oslo, April–May 2018). It includes 42 adults who stood on the floor for two ~8-minute sessions, one per playback device, counterbalanced, listening to silence and six rhythmic/EDM excerpts. They were asked only to stand, NOT to stand as still as possible, which distinguishes this record from the Championship of Standstill and means a quantity of motion from here is not comparable with one from those. Synchronised recordings: Qualisys infrared motion capture (22 markers, 200 Hz) Sweetzpot respiration belt (~10 Hz, participants S8–S42) Wii balance board (~54 Hz, irregular) Six Delsys body-worn accelerometers 1 Hz room-sound log There is also an extensive questionnaire battery (demographics, musical background, Big Five, empathy/IRI, music reward/BMRQ, genre preference) with a full codebook, and pre-computed quantity-of-motion summaries. Note on the balance board: its last two columns are load differences, not centres of pressure, and must be divided by the total load before use; see section 7.4 of the README for the conversion. This dataset underlies two articles by Zelechowska et al. (2020) in Frontiers in Psychology and Music & Science, as well as the doctoral dissertation behind them. It includes the original HPvSP analysis code. NOTE: copyrighted commercial stimulus audio is withheld from this record; a CC-BY acoustic-feature stand-in (features + spectrograms) is provided in sound_data/audio_features/ instead. Formats. tab-separated text (.tsv), plain text (.txt), PNG images and NumPy archives (.npz). Column dictionaries named *_dictionary.tsv define the columns, and the README describes any table that has none.