Abstract:
Stimuli used in timbre perception studies must be controlled carefully in order to yield meaningful
results. During psychoacoustic testing of individual timbre properties, (1) it must be ensured that
timbre properties do not co-vary, as timbre properties are often not independent from one another,
and (2) the potential influence of loudness, pitch, and perceived duration must be eliminated.
A mathematical additive synthesis method is proposed which allows complete control over two
spectral parameters, the spectral centroid (corresponding to brightness) and irregularity, and two
temporal parameters, log rise-time (LRT) and a parameter characterizing the sustain/decay
segment, while controlling for covariation in the spectral centroid and irregularity. Thirteen musical
instrument sounds were synthesized. Perceptual data from six listeners indicate that variation in the
four timbre properties mainly influences loudness and that perceived duration and pitch are not
influenced significantly for the stimuli of longer duration (2 s) used here. Trends across instruments
were found to be similar.
