Understanding sound radiation from the human voice has broad applications in room acoustical design, telecommunications, physical modeling of the singing voice, and virtual acoustics. Head simulators and head and torso simulators can provide simplified approximations to voice directivity, which motivate their use in room acoustical and other related measurements. Nonetheless, recent works have shown that scattering and diffraction due to the torso alter speech radiation patterns compared to those produced from an isolated head alone. Despite the improvements that including a torso provides, most commercial voice simulators neglect the effects of human legs. To better understand the impact of leg scattering and diffraction on voice directivity, this work presents measurements of a manikin with a head, torso, and legs. Comparing the results with those measured from human talkers shows that scattering and diffraction from human legs can impact voice radiation patterns, particularly above 1 kHz. The results also highlight the importance of high spatial sampling resolution when performing directivity measurements, as these scattering effects are easily spatially aliased in lower resolution sampling schemes.