The temporal evolution of spectral components is the primary cue for timbre recognition

In natural sounds, frequency components do not stay at fixed amplitudes — they change over time in characteristic ways, especially during attack and decay. Chowning’s key insight (citing Risset and Mathews) is that this spectral evolution is what listeners actually use to identify a sound’s timbre. Static or ‘fixed proportion’ spectra — the default of early synthesis — lack this evolution and produce the characteristic ‘lifeless’ electronic quality. This principle motivates FM synthesis: rather than statically summing partials (additive) or subtracting from a fixed waveform (subtractive), FM inherently produces spectra that change over time as the modulation index changes. Understanding this principle reframes synthesis goals: the target is not a static snapshot of a spectrum, but a trajectory through spectral states. The common misconception is that timbral character comes from the static harmonic content alone; in reality, how that content evolves is equally or more important.

Examples

Strike a piano key and listen: the attack contains many inharmonic transients that decay within milliseconds, leaving the sustained tone. The attack transient — not the sustain — is what makes a piano immediately identifiable.

Assessment

Given a synthesized tone and a natural instrument recording of the same pitch and duration, identify which spectral property (static harmonic content vs. spectral evolution during attack/decay) most accounts for the perceptual difference.