Quasi-harmonic musical instrument tones can be synthesized with various additive methods, but this approach requires a large number of parameters to describe the amplitude and frequency envelopes. Experienced users find it difficult to meaningfully manipulate so many parameters. A piecewise linear approximation with breakpoints reduces the data complexity. This study explores the perceptual implications of choosing the density of piecewise segments. Using a two-alternative forced-choice paradigm, listeners judged if the approximation was distinguishable from the original. Relative-amplitude spectral error and relative-amplitude critical-band error were found to be the best error metrics for predicting discrimination, accounting for about 80% of the discrimination variance. Strong correlations were observed between discrimination scores and the modified spectral incoherence based on the three strongest harmonics. Breath noise in the flute and bow noise in the violin appeared to cause increased discrimination issues.
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.