A statistical approach to analyzing sound tracings

  • Authors:
  • Kristian Nymoen;Jim Torresen;Rolf Inge Godøy;Alexander Refsum Jensenius

  • Affiliations:
  • Department of Informatics, University of Oslo, Oslo, Norway;Department of Informatics, University of Oslo, Oslo, Norway;Department of Musicology, University of Oslo, Oslo, Norway;Department of Musicology, University of Oslo, Oslo, Norway

  • Venue:
  • CMMR'11 Proceedings of the 8th international conference on Speech, Sound and Music Processing: embracing research in India
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper presents an experiment on sound tracing, meaning an experiment on how people relate motion to sound. 38 participants were presented with 18 short sounds, and instructed to move their hands in the air while acting as though the sound was created by their hand motion. The hand motion of the participants was recorded, and has been analyzed using statistical tests, comparing results between different sounds, between different subjects, and between different sound classes. We have identified several relationships between sound and motion which are present in the majority of the subjects. A clear distinction was found in onset acceleration for motion to sounds with an impulsive dynamic envelope compared to non-impulsive sounds. Furthermore, vertical movement has been shown to be related to sound frequency, both in terms of spectral centroid and pitch. Moreover, a significantly higher amount of overall acceleration was observed for non-pitched sounds as compared to pitched sounds.