Real-time bowing parameter sensing methods for articulated bow-string physical modelling

This paper presents bowing parameter estimation techniques developed specifically for real-time control of bowed-string physical models. By placing the sensors on the resonator instead of on the bow, the ability to seamlessly employ traditional bowing techniques is preserved. The sensing methods presented here are part of a larger project that aims to design and develop stage-ready virtual-acoustic bowed-string instruments.<br/><br/>The proposed approach utilizes load cells mounted beneath the string supports to sense bowing force and position in real time. An automatic offset adjustment algorithm is implemented to counter sensor drift, which enhances accuracy and long-term reliability. A Kalman-like feedback loop that implements level-dependent adaption of the bowing position was designed to avoid noisy estimates at small bowing forces. Experimental validation shows that this approach provides robust, usable estimates of bowing force and position with minimal calibration.<br/><br/>Bow velocity is inferred from string velocity data captured by an electromagnetic pickup system, utilising an approximation of the short-time probability density function. Preliminary bow velocity estimation results were obtained with an offline algorithm, showing promising capability for real-time application.<br/>