抄録
The purpose of our work is to design a real-time multimodal emotion extraction system based on fuzzy inference, and to implement it to a multi-core processor. We have been studied emotion extraction systems from voice and facial expressions separately. In general, emotions of human being are unstable therefore emotion extraction from single source will reach a limit. Multi-modal sensing like human beings do becomes important. We implemented a multi-modal emotion extraction system based on fuzzy inference to the Cell Broadband Engine TM. Our system deals with facial expressions and voice. To extend the system in the future, processes are partitioned by making thread for each source. The system can be extended easily by adding or modifying a thread. We got a good performance by assigning processing corresponding to each source to processor cores.
