音声認識手法による自動車の故障識別の試み

概要

論文の詳細を見る
Much of the examination and fault detection for automobiles have been made by human feeling. An operator beside an automobile listens to the operational sounds, that contain a lot of information about the state of the automobile, which the operator extract and recognize. In this study, instead of depending on human feeling, we employ the isolated word recognition technique in order to recognize the operation sounds. This method is as follows. In the training phase, various sound patterns are stored in the standard pattern file. In the recognition phase, the sound pattern is compared with the standard patterns by using pattern matching techniques, and the most similar standard pattern is selected. As a result, we can distinguish one engine from another, and any deviation in the number of revolutions through this system. This system shows the state of the automobile.
長野工業高等専門学校の論文