[3N3-J-10] Vision, speech: voice and communication

[3N3-J-10-02] Development of Open-source Multi-modal Interaction Platform for Social Experiment of Conversational User Interface

〇Akinobu Lee¹ (1. Nagoya Institute of Technology, Japan)

Keywords:Spoken Language Interaction, Spoken Dialogue System, MMDAgent

A development of a multi-modal interaction platform for diverse social experiment of conversational user interface is proposed. Recently simple spoken language interaction systems such as voice assistants have been rapidly in practical use, it is still necessary to elucidate various factors of rich interactions quantitatively based on wide variety of actual interaction data. The proposed system is based on a voice interaction building toolkit MMDAgent, adding some novel features to promote a testbed for social experiment and data collection of any speech interaction system on cloud environment. It includes facilities for system distribution and management, collection of interaction log and speech data, and easy connection with cloud-based chat system. The beta version of the software is available, and it will be released as open-source software to promote wider use for various speech-based conversational user interface.