Improving Auditory-Perceptual Assessment of Timbres: Testing Four Deconstructed Timbre Assessment Models

Introduction: Timbre is a central quality of singing and a much-discussed aspect of the perceptual aspects of sound. It remains a complex notion poorly understood in psychoacoustics, with no single variable or combination of variables consistently predicting timbre dimensions. Conceptually timbre varies on a colour continuum from dark to light extremes. These are strongly related to laryngeal and vocal tract adjustments. Perceptually, timbre assessment is influenced by spectral characteristics and formant frequency adjustments, though these dimensions are not independently perceived. Correlations between one singe acoustic measure and timbral changes are not established, with nuances of the timbre continuum eluding consistent perceptual assessment. A recent study demonstrated how acoustic predictive salience of voice category and voice weight across pitches contribute to timbre assessments and concludes that timbre may be related to as-of-yet unknown factor(s). This study tested four different conceptual models of auditory-perceptual assessments of timbre in progressively higher levels of detail; an intuition-based model, anatomical-analytical model, anatomical-analytical with auditory anchors, and an extended perceptual assessment model describing timbre as well as dimensions beyond timbre percepts.
Methods: Four separate auditory-perceptual panel assessments were conducted. Panel 1 and 2 involved 24 stimuli samples rated by 54 and 41 singing teachers with the intuition-based or anatomical model, respectively. Panel 3 involved 30 singing teachers rating 32 samples of stimuli (previous stimuli+additional stimuli for higher complexity) using the auditorily-anchored model, while panel 4 involved 12 expert singing teachers rating the same stimuli using the extended model. Inter-rater reliability scores were calculated by Fleiss Kappa (k).
Results: Panel 1 through 4 varied in overall accuracy and agreement. The intuition-based model showed overall 45% average accuracy (SD ±4%), k=0.289 (<0.001) compared to overall 71% average accuracy (SD ±3%), k=0.368 (<0,001) of panel 2. The auditory-anchored model showed overall 75% average accuracy (SD ±8%), k=0.54 (<0.001) compared with overall 83% average accuracy and agreement of k=0.63 (<0.001) for panel 4.
Conclusion: Timbre is more readily perceptually assessed along with other perceptual dimensions. The anchored and extended rating models performed with the highest accuracies and overall agreement, with the extended ratings showing consistently higher kappa values for all sub-dimensions. The intuition-based model performed comparatively the worst with the lowest reported accuracy and ICC. The study suggests new deconstructed approaches to assessing timbral and extended auditory perceptual dimensions.