The Role of FM-induced AM in Dynamic Spectral Profile Analysis
Résumé
(1) Recent work by Green and colleagues (cf. [Green, 1983]) has begun to demonstrate with
psychophysical methods that the human auditory system is capable of extracting a global
representation of the spectral envelope of a signal. This may serve to identify the resonance
structure characteristics of a sound source. However, this previous work has been exclusively
confined to relatively dense steady-state, inharmonic stimuli of simple spectral forms (a single
bump in the spectral profile). Dynamic stimuli, those with frequency jitter or vibrato, might be
helpful in reducing perceptual ambiguity in cases where there are not enough partials present in a
sound to clearly define a spectral envelope. They may do this by tracing out the spectral envelope
through time, thus increasing information about the resonance structure.
(2) While this seems intuitively obvious, previous work on high-pitched vowel identification in the
presence of frequency vibrato has yielded ambiguous results ([Sundberg, 1977]). These
researchers claim that intonation contours or vibrato had slight, or even detrimental, effects on
vowel identification. The present study demonstrates, to the contrary, that if the amplitude
behavior of a given partial is coupled with its frequency behavior according to a given spectral
envelope, this information can be used by the auditory system to discriminate and identify the
vowel quality or timbre of complex harmonic sounds.