Abstract
As the Internet of Things (IoT) continues to evolve, a growing number of wireless sensors have been integrated into daily life, posing new challenges to energy supply and communication. Point-to-point dynamic wireless power transfer and communication, enabled by user-based positioning and tracking services, hold great potential in the IoT. Here, we propose a speech-controlled reconfigurable intelligent metasurface (RIS) that translates natural-language commands into dynamically shaped electromagnetic beams by combining speech interaction, low-power RIS control, and a template matching algorithm, supporting real-time data communication and wireless power transfer for both static and moving targets. Unlike conventional RISs that rely on pre-defined control and external processing units, our approach provides the metasurface with visual and linguistic perception capabilities, enabling a paradigm shift from passive reconfiguration to active multimodal intelligence. The experimental results confirm the effectiveness of the speech-controlled RIS, while the measured results demonstrate that the RIS can provide a stable dc output that exceeds 4.61 V on dynamic targets. By enabling intuitive human-device interaction and effectively meeting the power supply requirements of small sensors, the proposed concept demonstrates strong application potential in IoT scenarios.