Clarifying spectral and temporal dimensions of musical instrument timbre

Authors

  • Michael D. Hall Department of Psychology, MSC 7704, James Madison University, Harrisonburg, VA 22807, United States
  • James W. Beauchamp School of Music, Dept. of Electrical and Computer Engineering, University of Illinois, Urbana-Champaign, IL 61801, United States

Keywords:

Experiments, Instruments, Law enforcement, Low pass filters, Musical instruments, Sensitivity analysis, Spectrum analysis, Speech recognition, Amplitude envelopes, Binary features, Higher harmonics, Instrument identifications, Multi-dimensional scaling, Musical instrument sounds, Relative amplitudes, Rise time, Spectral changes, Spectral dimensions, Spectral envelopes, Temporal cues, Temporal dimensions

Abstract

Classic studies based on multi-dimensional scaling of dissimilarity judgments, and on discrimination, for musical instrument sounds have provided converging support for the importance of relatively static, spectral cues to timbre (e.g., energy in the higher harmonics, which has been associated with perceived brightness), as well as dynamic, temporal cues (e.g., rise time, associated with perceived abruptness). Comparatively few studies have evaluated the effects of acoustic attributes on instrument identification, despite the fact that timbre recognition is an important listening goal. To assess the nature, and salience, of these cues to timbre recognition, two experiments were designed to compare discrimination and identification performance for resynthesized tones that systematically varied spectral and temporal parameters between settings for two natural instruments. Stimuli in the first experiment consisted of various combinations of spectral envelopes (manipulating the relative amplitudes of harmonics) and amplitude-vs.-time envelopes (including rise times). Listeners were most sensitive to spectral changes in both discrimination and identification tasks. Only extreme amplitude envelopes impacted performance, suggesting a binary feature based on abruptness of the attack. The second experiment sought to clarify the spectral dimension. Listener sensitivity was compared for a) modifications of spectral envelope shape via variation of formant structure and b) spectral changes that minimally impact envelope shape (using low-pass filters to match the centroids of the formant-varied envelopes). Only differences in formant structure were easily discriminated and contributed strongly to identification. Thus, it appears that listeners primarily identify timbres according to spectral envelope shape. Implications for models of instrument timbre are discussed.

Downloads

Published

2009-03-01

How to Cite

1.
Hall MD, Beauchamp JW. Clarifying spectral and temporal dimensions of musical instrument timbre. Canadian Acoustics [Internet]. 2009 Mar. 1 [cited 2021 Nov. 29];37(1):3-22. Available from: https://jcaa.caa-aca.ca/index.php/jcaa/article/view/2111

Issue

Section

Technical Articles