A self-powered patch read silent speech and controlled games and avatars

The flexible patch recognized 15 silent-speech categories and was used for visual, gaming and virtual-reality communication tools.

Editorial illustration — not from the study.

The system combines a flexible electronic-skin patch based on triboelectric materials with a multilayer artificial-intelligence model. The patch records mouth movements without requiring spoken sound, while the model interprets the signals as silent speech.

What the patch recognized

The researchers characterized the patch’s structure and electrical performance and compared mouth movements during spoken and silent speech. They reported strong similarities between the two types of movement in their timing and frequency patterns. A model trained on 15 word categories, with 200 samples per category, reached a reported accuracy of 98.75%. The team also demonstrated three applications: real-time visualization of silent speech, an orofacial-controlled version of the game “Feeding Frenzy,” and a virtual-reality communication platform in which avatars could hold dialogues.

Why silent speech matters

The patch offers a hands-free approach to communication that does not depend on producing an audible voice. The researchers say it could support communication for people with laryngeal or lingual impairments, as well as human–machine interfaces, assistive communication systems, virtual-reality communication and biomedical rehabilitation.

Evidence and open questions

This is a journal research article reporting sensor characterization, a comparison of spoken and silent mouth movements, machine-learning classification and demonstrations of three applications. The abstract does not give the number of participants, explain how training and testing samples were separated, or report performance outside the 15 categories used for training. It also does not establish how well the system would work across different users, environments or longer conversations.

A new tool estimates the forces on feet from movement data alone

Researchers developed OpenGRF, an OpenSim-based tool that estimates forces, turning effects and pressure under the feet from joint movement data. In tests with seven healthy adults, it matched force-plate measurements most closely for vertical forces and ankle moments, while some sideways forces and free moments were harder to predict.

1 day ago·10 views

Engineering & Technology

Tiny lead crystals can produce laser light after a moisture-driven change

Researchers report that moisture-driven changes in cesium lead bromide crystals can produce 300–700 nanometre crystals capable of lasing individually. The result suggests that miniature light sources may not need regular, micrometre-scale cavities.

2 days ago·48 views

Engineering & Technology

Perovskite films enable long-wave infrared imaging with silicon cameras

Researchers have demonstrated long-wave infrared imaging with solution-processed two-dimensional perovskite films and a silicon-based camera. The centimeter-scale imagers detected 9.5-micrometer infrared light with sensitivity down to about 20 pW per square micrometer.

2 days ago·19 views