How a new device can use static electricity on your clothes to help you chat to Siri or ChatGPT

Embargoed until: Publicly released: 2025-10-10 10:30

International

Photographic image showing the A-T extile–integrated garment for voice perception. Scale bar, 4 cm. Credit: AAAS

Your clothing could help you chat to AI systems, according to international researchers, who have developed a new textile that amplifies the static electricity generated in your clothes when you speak. The device has an accuracy of up to 97.5% for voice perception and was able to distinguish subtle sounds while maintaining accuracy in loud environments. The team were able to use it to control household smart appliances like lights and air conditioners, interface with Google Maps, and ask ChatGPT for cocktail recipes and a travel itinerary. The team say their developments lay the foundations for the next generation of intelligent, wearable voice-AI systems.

Media release

From: AAAS

Textile harnesses static electricity on clothing to help AI listen to voice commands
Science Advances

A new type of textile helps AI recognize voice commands by amplifying electrostatic charges generated by speaking. When added to fabric, the washable and lightweight AI-textile demonstrated an accuracy of up to 97.5% for voice perception, enabling it to control household appliances remotely.

“The integration of imperceptible voice AI systems into clothing will provide easy access to AI, while providing new application prospects in various areas such as healthcare, fitness monitoring, and personalized assistance,” Beibei Shao and colleagues write.

“This work lays the foundation for the next generation of intelligent, wearable voice-AI systems.”

Talking to generative AI models, including ChatGPT, provides an AI interaction that feels more like a phone call than a search engine interaction. Yet, acoustic devices that support talking to AI are generally unwieldy and hard to carry in daily life. Now, Shao et al. present a portable, unobtrusive, flexible, and washable textile device that senses and amplifies electrostatic charges on fabric generated from speaking.

The graphite-like carbonized textile has a coating of 3D tin-sulfide nanoflowers enmeshed in silicon rubber. In tests, it showed the ability to distinguish subtle sounds while maintaining accuracy in loud environments. When integrated with a deep learning model, the AI-textile reacted to voice requests with an accuracy of 93.5 to 97.5%, depending on the activity. It connected with “internet-of-things”-enabled smart home appliances to complete 10 commands, including turning the air conditioning and a lamp on and off.

It also interfaced with Google Maps, responding to requests for navigation to national parks and temples. The authors also successfully tested the AI-textile’s ability to engage with ChatGPT by asking for cocktail recipes and a travel itinerary.

Journal/
conference: Science Advances

Research:Paper

Organisation/s: Soochow University, China

Funder: This work was supported by the following: National Science and Technology Council, Taiwan grants 113-2628- E- 005- 003, 113-2112- M- 005- 004, 113-2221- E- 007- 038, and 113-2112- M- 005- 011 (to Y.-C. L.); “Innovative Center on Sustainable Negative-Carbon Resources” from The Featured Areas Research Center Program within the framework of the Higher Education Sprout Project by the Ministry of Education, Taiwan (to Y.-C .L.); National Natural Science Foundation of China grants 62274116, 52103306, and 22408248 (to B.Su., R.L., and B.Sh.); 2024 China-France Hubert Curien Partnership Program-C ai Yuanpei Project CFPHC2024006 (to B.Su. and B.Sh.); Natural Science Foundation of Jiangsu Province grants BE2022025 and BK20210719 (to B.Su. and R.L.); and The 111 Program and the Collaborative Innovation Center of Suzhou Nano Science and Technology (to B.Su. and B.Sh.).

Media Contact/s

Contact details are only visible to registered journalists.