Intuitive Gait Modeling using Mimetic-Words for Gait Description and Generation

Gait is one of the most familiar action for us, that is why we can distinguish slight difference of human gaits and perceive their impressions. However, the relationship has been never explored because of the absence of intuitive labels for the slight differences. In this paper, to solve this problem, we propose a intuitive gait model using Japanese mimetic-words. A mimetic-word has sound-symbolism, which means that there is an association between linguistic sounds and sensory experiences, and the phonemes of a mimetic-word is strongly related to the visual sensation. Thanks to the sound-symbolism, Japanese mimetic-words have a possibility of modeling gaits intuitively. Thus, we have previously proposed a method which describes gait with a mimetic-word. In this paper, in the opposite direction, we propose a method which generates gait from a mimetic-word, and confirm the effectiveness of the proposed intuitive gait model which consists of the phonetic-vector through evaluations of both the generation task and the description task.