Basic Study on Visualization of Environmental Sound Using Smart Glass to Support Hearing-impaired Persons

The hearing-impaired persons are quite hard to catch the environmental sounds, which includes various kinds of information that is needed to know what is happening in surrounding environments around them. For example, some hearing-impaired persons cannot take in laundries due to the inability to catch the sound of rain, or encounter a scene that is dangerous to themselves without noticing the approaching sound of the vehicle. For such a reason, it is of great importance to support the hearing-impaired persons with notifying them various kinds of events with visual to measure. The purpose of this research is to identify various kinds of environmental sounds that can be important clues from the viewpoint of safe life by using machine learning, and display them on the smart glass. The acoustic features of the environmental sounds are extracted from their time-, frequency- and time-frequency-domain characteristics, and the kind of them are identified based on the extracted features. In addition, by using the smart glass , the hearing-impaired person can catch the environmental sounds without obstructing their field of view. In this paper, the validity of the basic study on environmental sound classification by using machine learning is discussed.