Language is the most human of all interfaces. Therefore, the moment people interact with voice assistants, they interpret a "personality" in their counterpart. What possibilities do we have to describe and shape this “personality"? What do we have to pay attention to in order to design a good voice user interface? How can we design such a character in such a way that it convinces in the usage context and transports the brand values coherently?
While working on different speech assistance and speech recognition systems, I have always dealt with these questions. The result was a lecture that I was able to give at the TU Dresden or at the wearable conference Wear It Berlin. The two most important findings were that such a personality can easily be prototyped with speakers, but that one should always be aware of the uncanny valley effect. A speech interface should not seem too human, because - especially in western cultures - this is quickly perceived as unpleasant.