This paper presents a tracker-based hand gesture recognition system designed for robust, real-time music player control on Android devices, leveraging precise hand landmark tracking and rule-based gesture modelling. In contrast to recurrent neural models, which exhibited segmentation artifacts and false positives in prior work, the proposed approach achieves higher reliability and responsiveness with lower device resource consumption. A comprehensive literature survey is provided, contemporary systems are reviewed, and our tracker-based method is analytically and empirically compared to a custom LSTM-based sliding window approach. The results show that tracker-based techniques represent a strong alternative for embedded gesture-driven applications.