| Summary: | 碩士 === 長庚大學 === 資訊工程學研究所 === 96 === In recent year, there are more and more popular with computer science and music. People use computers to compose music, but it lacks of interactive. In this thesis, we use three dimensional acceleration sensor for developing a conductor system. It provides us a natural way to wave our arm to control songs, just like a real concert occasions.
In this thesis, we provide a strategy of Beat Detection and Prune, which analysis signals and will be able to get the beat interval. According to the beat interval, we can calculate the tempo and recognition beat type. We decided upon was simple but effective: gesture speed is mapped to tempo, and gesture size is mapped to volume.
In this research, we collected some beat types: two beat, three beat, four beat, and six beat. The total of conducting data is about 2000, and the average of recognition rate is 92.9%. Besides, in order to evaluate the correct rate of Beat Detection, we collected 1800 testing data. We found that the average error of our system is about 0.0296 seconds to reach 90% correct rate.
|
|---|
