For many people, not being able to speak is a significant problem. His 2014 study on voice disorders in the United States found that: Approximately 18 million adults One group had difficulty speaking using their vocal tract, while more than half of that group experienced debilitating language impairment for more than a week.
New non-invasive, self-powered wearable devices offer one potential way to address this medical need. The technology consists of a lightweight patch placed on a person's neck that measures the movement of the person's neck. An off-device processor then converts these signals into audio, which is then played in place of the human voice.
“Just as this material converts muscle movement into electricity, it can also induce electricity as an input signal to mechanical vibrations in devices that produce sound.” —Jun Chen, UCLA
research teamA research team led by Jun Chen, assistant professor of biomedical engineering at UCLA, has developed a flexible, electromagnetically responsive wearable that measures the subtle movements of neck muscles. A downstream device, rather than a self-powered sensing patch, then decodes the muscle movements sensed by the patch into speech using a machine learning algorithm trained to recognize a configured phrase dictionary. .
This technique is useful for people who have injuries or illnesses that cause vocal cord paralysis, as well as people who have undergone surgery such as laryngectomy (removal of the larynx, which contains the vibrating muscles) to remove some or all of the vocal cords. may also bring benefits.
However, device performance is limited by the number of sentences that can be played. The current prototype AI model simply selects from her five sentences: “I hope the experiment works!” “Merry Christmas!”; “I love you!”; and “I don't trust you.” Trained and tested based on neck movements. Subjects were asked to move their heads as if they were talking without making any vocalizations.
Medical technology options for the voiceless
The new device works in conjunction with other medical technologies to treat vocal cord disorders, such as artificial larynxes and voice generators that vibrate the neck to recreate the movement of lost vocal cords. The UCLA patch approaches this problem in a different way, converting the user's silent neck muscle activity into computer-generated audio.
Dr. Barbara Messing, a clinical educator at the medical device company Atos Medical who is not affiliated with the UCLA project, said this new approach is a welcome addition to the field of voice assist devices. It’s better,” Messing said. “Voice prosthesis is the gold standard for patients who have had a laryngectomy, but it is not the case for all patients. Having more options for patients can only improve their quality of life. That’s what we all want.”
To make the user's unspoken voice audible, the device passes the evoked muscle movement signals to a machine learning model running on an external processor. The model is trained to detect patterns of muscle movements that correspond to a certain number of predefined sentences. When it detects one of these phrases, the processor vibrates the patch's surface as a speaker to play the sentence. “Just as this material converts muscle movements into electricity, it can also direct electricity as an input signal into mechanical vibrations within the device, producing sound,” Chen says.
Researchers have developed an AI-powered voice patch that is applied non-invasively to the throat. A thin magnetic induction coil and the flexible magnetic material inside it work under the skin to infer neck and throat movements. Jun Chen Lab/UCLA
Creating a patch
The system's throat patch applies new materials science research from Chen's group that exploits a property called magnetoelasticity, in which the strength of a material's magnetic field changes as it stretches or compresses. Regular daily activities and neck movements cause the patch to stretch, creating changes in the magnetic field that are measured by a built-in flexible induction coil. These materials work together to passively sense the minute 3D movements of the user's neck muscles.
Magnetoelectric effects have been observed in metallic materials since the 19th century, but the stiffness of such materials has made biological applications, such as measuring contraction and expansion of human neck muscles, difficult. Chen's group has discovered a way to make highly stretchable materials magnetoelastic by embedding particles of harder magnetoelastic material into sheets of flexible polymer.
The flexibility of the new magnetoelastic material allows it to adhere closely to the user's neck muscles and precisely track their movements in a way that was not possible with similar sensors made from previously known magnetoelastic materials. I can. To further increase sensitivity, the researchers formed the material into a kirigami pattern (an origami-like paper craft that can be cut), allowing the sensor to operate evenly with smaller stretches and deflections.
Going forward, the group will work on applying its research to medical devices, Chen said. “In the future, we want to optimize the device and standardize the manufacturing procedure for mass production,” Chen says. “We need to improve the software and hardware to increase the translation vocabulary and accuracy, and make the devices more user-friendly. All these issues need to be solved one by one,” Chen said in the study. He estimates the group will be able to produce a working medical device “within three to five years.”
The researchers published their findings earlier this month in the journal Nature Communications.
From an article on your site
Related articles on the web
