Cassie Mitova

Research Projects: Robotic Voices

Synthesized Voices and the Uncanny Valley

Does the likability of an artificial voice depend on how much it sounds like a human?

Artificial voices have become more and more human-like since their digital conception in the 1980s. It has been said that some “synthesized” voices are barely distinguishable from that of a real human. However, something sounding more human-like may not necessarily be appealing to a real human being.

Masahiro Mori’s Uncanny Valley theorem plots people’s reactions to robots that range from looking completely inanimate to being a perfect recreation of a human. The theorem states that there is a point at which a robot looks almost human, but not quite, and instills revulsion in human observers. This study will determine whether the appeal of synthesized voices follows the same pattern as the appeal of robots—the Uncanny Valley theorem—or something different entirely.

Participant responses to this poll will be used to create a graph comparing human-likeness to appeal in a similar setup to Mori’s graph. A comparison of the two will reveal whether an uncanny valley exists for synthesized voices like it does for robots.