After a stroke, some people may be able to move their lips to form words, but may not be able to speak clearly enough for others to understand.

A soft device worn around the neck aims to convert these silent efforts into fluent spoken sentences, with the potential to facilitate everyday communication and restore some independence in daily care.

listening collar

Revoice is a soft, flexible choker worn around the neck that detects physiological signals that would normally never be heard.

Developed by engineers. cambridge universitythis device is designed with comfort, washability, and everyday use in mind.

The study was led by Dr. Luigi Occhipinti. His lab specializes in wearable sensors that convert subtle body signals into meaningful outputs.

Early tests suggested the idea could deliver liquid speechbut the validity of that experience will be determined by more extensive testing.

When the language muscle misfires

Many stroke survivors live with slurred speech or slow speech. dysarthriadriven by weak or uncoordinated muscles.

Even if you have prepared the right words, your spoken words may sound rushed, quiet, or choppy. Recent reports describe deep frustration in patients when clear thinking does not cooperate.

“People who suffer from dysarthria after a stroke can be very frustrated because even though they know exactly what they want to say, it’s physically difficult to say it because the signals between their brain and throat have been disrupted by the stroke,” Dr. Occhipinti said.

Signals from the throat and pulse

In this clinical study, volunteers silently formed short phrases in controlled sessions while wearing a collar.

Sensors track small vibrations in the throat and carotid pulses (heart waves in the neck arteries) to add context.

The collar used a fabric strain sensor, a fabric strip that changes resistance as it stretches, so it remained soft against the skin.

Because neck movements are also caused by swallowing and rotation, separating vocal patterns from everyday noises is an important engineering constraint.

Decoding the movements of the silent mouth

Rather than listening to sounds, Revoice interprets silent words, or words uttered without an audible voice, from muscle activity.

The software splits the signal into fast time slices and predicts words Play the pieces to the throat pattern that unfolds in real time.

Continuous decoding avoided forced pauses between words, a common problem when systems rely on fixed windows for recognition.

Training remained important because each person produces slightly different movement signals as fatigue and recovery vary from day to day.

When context fills the gap

The system did more than just map movements to words, since short snippets rarely convey enough meaning. every day talk.

The model read the timing of the pulses, labeled them with basic emotions, and combined them. time Or the weather leads to expansion.

These signals guided a large-scale language model that transformed a few spoken words into complete sentences. Although the user triggered the expansion with a simple nod signal, automatic rewriting must be faithful to the speaker’s intent.

Treatments rarely match the real story

Most clinics offer speech therapy that includes: dysarthria Rely on repeated practice to rebuild clearer mouth movements.

Patients can work with therapists to learn scripted words, but everyday conversations require quick timing, breath control, and steady voice volume.

Patients usually learn to cope with repetitive training after some practice, but often have difficulty with open-ended questions and everyday conversations.

Wearable tools that speak in real time have the potential to support everyday conversation practice, but it depends on training and daily practice.

What we learned from the initial trial

In tests with five patients and 10 healthy volunteers, the word error rate, which is how often words were incorrectly recognized, reached 4.2%.

Only 2.9% of the sentences were answered incorrectly, and participants reported a 55% increase in satisfaction.

These numbers come from small vocabularies and small cohorts, so they still cannot predict performance in different environments.

Even with its limitations, immediate feedback can be useful for practicing conversational rhythms while concurrently continuing speech therapy.

Many assistive tools are augmentative and alternative forms of communication that bypass voice with other inputs.

Switch and eye-tracking spells are reliable, but messages arrive slowly. conversation Moves fast.

Brain implants can help some people who can’t move at all, but surgery and clinical support are rare.

Neck devices fit into another intermediate group, but they still may not be usable if you have poor head control or severe paralysis.

Giving voice to stroke patients

In the United States, hindrance As post-stroke symptoms remain common, the demand for practical communication aids remains high.

Large-scale trials will require awkward real-life tests such as coughing, laughing, and changes in posture that can distort the sensor.

The researchers suggest future versions that would work offline and in more languages, reduce data sharing, and expand who can use the data.

Until these steps are completed, collars will continue to be a viable tool for most families, rather than something that can be purchased off-the-shelf.

Revoice has shown that by combining soft sensors and language software, silent mouth movements can become clear speech in real time.

If extensive testing confirms its reliability and protects privacy, the device could help people train their clinics with real conversations.

The research will be published in a journal nature.

—–

Like what you read? Subscribe to newsletter We bring you fascinating articles, exclusive content, and the latest updates.

Please check it out earth snapThis is a free app provided by. Eric Ralls and Earth.com.

—–