Listener preferences for vibrato rate and extent in synthesized vocal samples

Abstract

An experiment was devised to test the preferences of singers, vocal teachers and choral directors for vibrato rate and extent on the /a/ and /i/ vowels and with two different “gender” singers by using the voice synthesis program Madde, which is capable of varying vibrato rate, extent, fundamental frequency, and formant frequencies. The study also sought to determine if the vowel and “gender” of the singer (as determined by the formant frequencies used for the simulation) have an impact on listener preferences. Subjects answered a series of questions about their musical training and occupation, then listened to four samples each, two of the “male” voice performing the vowels at 220 Hz, and two of the “female” voice performing the same vowels at 440 Hz. The order in which the samples were presented and the initial vibrato rate and extent conditions of the simulations were varied to try to negate any order effects on the data. The investigator adjusted the vibrato rate and vibrato extent until the listener felt his or her preference was being best represented. Listeners displayed a wide variety of preferences, with those most involved in musical theater and early music performance preferring a narrower vibrato extent.