Study on Aiding Systems for the Hearing Impaired : Discrimination of Alarm
Kazufumi Suzuki, Takashi Sasaki, Tomoo Takeguchi, Minako Ohashi · The Proceedings of Conference of Kansai Branch · 2003
Hearing aids could help the hearing impaired to carry conversations and watching TV. However, once outdoors, hearing aids used commonly causes the discomfort to many. It is very common not to use hearing aids outdoors though it may increase the danger to miss alarm or siren. So, it is important to develop an aiding system which could notice the hearing impaired not only by hearing but other manners, such as vibration, lighting and so on. Then, the sudden alarm could be notified so that the safety of the hearing impaired could be improved. As the first step to develop these aiding systems, sirens for emergency vehicles, and alarms for railroad crossing were used for tests of discrimination and notice. The sounds of these alarms are analyzed by FFT and discriminated by band pass filter. However, wrong discrimination occurs when the noise exists inside of bands. The neural network was used to improve discrimination for various alarm sounds. The results of discrimination did not change because the classification of the sounds was by FFT. The learning by the neural network will be more effective to aiding system after alarm is characterized with time and frequencies.