 This study presents a fully integrated, standalone stretchable device platform that wirelessly measures diverse vibrations and muscle electrical activities from the throat through machine learning-based analysis. The modified composite hydrogel with low contact impedance and reduced adhesion provides high quality long-term monitoring of local muscle electrical signals, while the integrated triaxial broadband accelerometer also measures large body movements and subtle physiological activities of vibrations. The combined data processed by a 2D-like sequential feature extractor with fully connected neurons facilitates the classification of various motion speech features at a high accuracy of over 90% which adapts to the data with noise from motion artifacts or the data from new human subjects. The resulting standalone stretchable device with wireless monitoring and machine learning-based processing capabilities paves the way to design and apply wearable skin-interfaced systems for the remote monitoring and treatment evaluation of various diseases. This article was authored by Hong Chengxiu, Weihao Zheng, Yang Zhong and others. We are article.tv, links in the description below.