A model of the auditory process is proposed for performing singers, which incorporates the added signal from bone conduction, as well as the psychological distance for subjective preference of the performer from the acoustic sound field of the stage. The explanatory power of previous scientific studies of vocal stage acoustics has been limited by a lack of an underlying theory of performer preference. Ando’s theory, using the autocorrelation function (ACF) for parametrizing temporal factors, was applied to interpretation of singer sound field preference determined by the pair comparison method. Melisma style singing (no lyrics) was shown to increase the preferred delay time of reflections from a mean of 14 ms with lyrics to 23 ms without (p<0.05). The extent of the shift in preferred time delay was shown to be directly related to minima of the effective duration of the running ACF, (τe)min, calculated from each singer’s voice. Voice matching experiments for singers demonstrated a strong overestimate of the voice outside the head compared with the singer’s own voice (22.4 dB overestimate, p<0.01). Individual singer melisma singing delay preferences were compared for “ah” versus “hum” syllables, and the increased delay preference (41 ms) was shown to be correlated with (τe)min (r2<0.68, p<0.01). When the proposed bone conduction model was applied, using the measured overestimate of sound level of the singer’s own voice for each singer (9.9 dB mean overestimate difference between “ah” and “hum,” p<0.01), the relationship of singer preference to (τe)min was improved (r2 = 0.97, p<0.01).
Thesis advisor: Yoichi Ando
Copies of this thesis are available from the author by inquiry at BRC Acoustics, 1741 First Avenue South, Seattle, WA 98134 USA. E-mail address: firstname.lastname@example.org