This article presents a self-powered, wearable sensing-actuation system based on soft magnetoelasticity, designed to enable assisted speaking without relying on vocal folds. The system consists of a lightweight (7.2 g) and skin-like (7.83 × 10^5 Pa) sensing component that captures extrinsic laryngeal muscle movements and converts them into high-fidelity electrical signals, and an actuation component that translates these signals into voice signals. The sensing component features a kirigami structure for enhanced sensitivity and stretchability, allowing it to detect muscle movements in all three dimensions. The system demonstrates high sensitivity (94.68% accuracy) and quick response time (40 ms), and it is intrinsically waterproof. The device can produce clear voice output with an SPL of over 40 dB, even under heavy perspiration, and has a maximum stretchability of 164%. The system's effectiveness is validated through experiments involving eight participants, showing consistent performance and user adaptability. This technology offers a promising solution for patients with voice disorders, facilitating communication during recovery and enhancing their quality of life.This article presents a self-powered, wearable sensing-actuation system based on soft magnetoelasticity, designed to enable assisted speaking without relying on vocal folds. The system consists of a lightweight (7.2 g) and skin-like (7.83 × 10^5 Pa) sensing component that captures extrinsic laryngeal muscle movements and converts them into high-fidelity electrical signals, and an actuation component that translates these signals into voice signals. The sensing component features a kirigami structure for enhanced sensitivity and stretchability, allowing it to detect muscle movements in all three dimensions. The system demonstrates high sensitivity (94.68% accuracy) and quick response time (40 ms), and it is intrinsically waterproof. The device can produce clear voice output with an SPL of over 40 dB, even under heavy perspiration, and has a maximum stretchability of 164%. The system's effectiveness is validated through experiments involving eight participants, showing consistent performance and user adaptability. This technology offers a promising solution for patients with voice disorders, facilitating communication during recovery and enhancing their quality of life.