Although we can sense someone’s vocalizations with our ears, nose, and haptic sense, speech is invisible to us without the help of technical aids. In this paper, we present three interactive artworks which explore the question: “if we could see our speech, what might it look like?” The artworks we present are concerned with the aesthetic implications of making the human voice visible, and were created with a particular emphasis on interaction designs that support the perception of tight spatio-temporal relationships between sound, image, and the body. We coin the term in-situ speech visualization to describe a variety of augmented-reality techniques by which graphic representations of speech can be made to appear coincident with their apparent point of origination.