1.4 Conclusion

Outlines of HARK 1.0.0 have been reported as above. Sound source localization, sound source separation and recognition of separated sounds, which are the fundamental functions for understanding sound environment, have been realized with the middleware FlowDesigner as modules and its application to ears of a robot has been reviewed. HARK 1.0.0 provides functions to develop furthermore the researches on robot audition. For example, functions for sound source processing, detailed set-up functions for various parameters of sound source separation and setting data visualization / creation tool are provided. Moreover, support for Windows and interface for OpenRTM are also in progress. Although recognition is realized at some level only by downloading and installing HARK, if performing the tuning for shape and usage environment of an individual robot, even better performance of source localization, sound source separation and recognition of the separated sound would be obtained. For exposing such know-how, it will be important to form HARK community. We hope this document provides users with an opportunity to exceed the critical mass of robot audition R&D researchers.