This paper presents a silent speech recognition system (SSRS) using tattoo-like electronics and machine learning. The system utilizes imperceptible tattoo-like electrodes attached to the face to record high-quality surface electromyographic (sEMG) signals. A machine-learning algorithm, deployed on a cloud server, accurately recognizes the silent speech. Experiments show high accuracy (92.64%) in recognizing 110 daily words, even with significant facial deformation. The SSRS successfully functions in various real-world scenarios, including noisy and dark environments.
