Design and development of high sensitivity sensor for finger tracking in piano playing

The art of planning the movement of hands in order to produce the desired sound of the piano is one of the important part of piano technique. Various researches have attempted to unveil the technique of virtuoso pianists using technologies. These researches employ different types of sensors in order to capture motion data in piano playing. However, one area in these researches have been under-represented, which is the finger position of a musician while playing the musical instrument. In piano technique, it is very important to study the finger position that could land on any position along the surface of one single key. Researches that embark on this area face a common problem, the sensors used in these researches are short in range and directly in contact with the pianist, which causes a change of piano playing experience. To avoid obstruction to the pianist, the finger tracking sensor should be placed away from the keyboard and track human finger across a set of wooden piano keys with 13 cm thickness. However, current technologies of capacitive finger tracking emphasizes either on near proximity sensing or remote gesture recognition, where both of them do not focus on accurate remote positional tracking that is required in this application. The range limitation of the current sensors is mainly affected by circuit design and size of electrodes. Concluding the challenges, a high sensitivity finger tracking sensor is proposed. A series of researches and comparing are carried out to select a high sensitivity capacitive sensing method with low noise. Upon selecting the sensing method, circuit modifications and components parameters tuning are applied. This includes minimizing stray capacitance and filtering noise. In addition to this, the sizes and arrangements of electrodes are optimized where electrodes are coupled with wooden keys to increase sensitivity of the sensor. The final signals from the sensor are digitalized and trained using artificial neural network to obtain positional data. This prototype sensor is developed to track different position of the fingers on five keys of the piano. To validate the design, 500 sets of independent input data with known output position were used to test the network. The output shows that the average error between the test set and the desired target is 7 mm, which translate to 83.79% accuracy. The error and accuracy from the output is reasonably good for all data set. To summarise, this research presents the design and development of a high sensitivity finger tracking sensor through sensor hardware design and signal processing. The outcome from this work could be applied to researches that require detection of minute capacitance induced by human finger