Objective It is generally quite easy to recognize and identify familiar of people by just hearing their voices. In some legal contexts, the voice has even been used as a tool for speaker identification. Still, voice mimicry can represent a problem. In the past, some studies of mimicry have been published revealing that imitators manage to replicate some, but rarely all voice characteristics. Typically, similarities have been found regarding dialect, pitch and timing patterns and also formant frequencies (see e.g., Eriksson & Wretling, 1997). These features are varied considerably and systematically in speech production. However, also voice source properties can be varied within wide limits, even though its control is mostly unconscious. This study aimed at including also source parameters in comparisons between target and imitation of voices.

Methods A female actor (co-author RM), professionally specialized in voice mimicry, volunteered to produce imitations of seven well-known voices recorded from the Internet. The actor repeated the text of the target voices’ running speech (I) with her own conversational voice; (II) while imitating the target voices, and also (III) produced sustained vowels while mimicking the target voices. The running speech material was analyzed by long-term-spectrum and, for formant frequencies, by PRAAT. Voice source data were obtained by inverse filtering the sustained vowels and from the LTAS H1-H2 parameter.

Results Preliminary results suggested that while imitating the target voices the actor varied formant frequencies and pitch and timing patterns, as expected. In addition, she varied voice source characteristics within extreme wide, particularly, and independent of pitch, the closed quotient. The variations reduced the differences between actor’s normal voice and the target voices in at least some of the analyzed parameters.

Conclusions In voice mimicry, great deviations from the imitator’s typical voice characteristics can be observed, which decrease differences in resonatory and phonatory voice characteristics between the imitator and the target. Studying voice characteristics in professional mimicry seems a promising avenue to a deeper understanding of controllable voice variability.