samtalets prosodi/prosody in conversation

 
 

We have formulated a visionary goal: to learn enough about human face-to-face interaction that we are able to create an artificial conversational partner that is human-like, or as Cassell (2007) puts it, ”acts human enough that we respond to it as we respond to another human”.


Note that this is not a case of implementing science fiction – it is about testing scientific hypotheses. Paraphrasing the methods used in cognitive science, we implement models of human behavior and put these to the test in interactions with humans. This is how we validate our understandings.


While our dream of human-likeness may eventually turn out to be overly ambitious, it has already been useful to us by guiding our research efforts towards black holes in our knowledge about human conversation. It is for example evident that state-of-the-art speech technology neither sounds like a conversational partner, nor understands fundamental aspects of human conversational behavior.


A massive amount of cross-disciplinary research is needed to realize the visionary goal. Although we have a long way to go, we learn useful lessons from every step. The speech group at KTH Speech, Music and Hearing is currently involved in a half-dozen national and international research projects orchestrated towards our goal, and with a common research focus: interactional aspects of spoken face-to-face communication.


This site describes the project Prosody in conversation (Samtalets prosodi).

 

Visionary goal

We have formulated a visionary goal: to learn enough about human face-to-face interaction that we are able to create an artificial conversational partner that is human-like.