Interval and Ratio Scaling of Spectral Audio Descriptors.

Area of Science:

• Psychoacoustics
• Auditory perception
• Signal processing

Background:

• Accurate psychophysical scales are crucial for understanding auditory perception.
• Previous research often used complex, multivariate stimuli, limiting causal inference.
• Independent control over audio descriptors is needed for precise scale derivation.

Purpose of the Study:

• To derive psychophysical scales for key audio descriptors: spectral centroid, spread, skewness, odd-to-even harmonic ratio, deviation, and slope.
• To establish reliable methods for quantifying sensation magnitudes of these audio features.
• To enable direct cause-and-effect relationships between audio descriptors and perceptual dimensions.

Main Methods:

• Conducted two experiments using partition scaling methods with synthesized, independently controlled audio stimuli.
• Experiment 1: Listeners estimated relative differences between successive audio descriptor levels (interval scale).
• Experiment 2: Listeners adjusted stimuli to create perceptually equal intervals (ratio scale).

Main Results:

• Listeners could estimate intervals, but interval scale reliability varied with stimulus spacing.
• Ratio scales derived in Experiment 2 demonstrated excellent reliability across all tested audio descriptors.
• Empirical units were assigned to ratio scales for cross-descriptor comparisons.

Conclusions:

• Psychophysical scales for audio descriptors can be reliably derived using univariate stimuli, allowing for causal analysis.
• The study provides a robust understanding of how sensation magnitudes for multiple audio descriptors are perceived.
• This research advances the field by offering precise, quantifiable relationships between acoustic features and auditory experience.