A Computer Generated Emulation of a subject

摘要

A system for emulating a subject in the form of a computer generated talking head to allow a user to interact therewith. The system comprises a processor 3, user interface (GUI) and a personality storage section 7 containing information created by or about the subject, preferably from social media websites, emails. The GUI displays 19 a talking head which comprises the subjects face and appears to output speech from the mouth of the face. The GUI includes a receiver for receiving a query 15 from the user and the talking head is configured to respond to the query. A dialogue section 4 of the processor generate a response to the query inputted by the user, the response is created from the information contained in the personality storage section. The response is converted 5 into a sequence of acoustic units and a statistical model is used to determine probability distributions which correlate an acoustic unit to an image vector and speech vector. The output 17 comprises a synchronized sequence of speech and image vectors which appears to make the GUI head talk. Preferably a set of rules are applied to the dialogue section. Expressions 6 may also be included into the response outputted. Alternatively the output may be in the form of text and the response being derived from converting the query into a word vector, as shown in fig.5b.