Speak without sound: This AI system turns muscle movements into speech

AI is already changing the medical sector today – primarily through chatbots that users ask about health topics. But AI can do much more for people with physical limitations. This is now also shown by a team of researchers from the University of Science and Technology in Pohang. They tried to make it possible for people without voices to speak again.

AI analyzes muscle movements

As Digitaltrends reports, the researchers have developed a gadget that analyzes the neck muscles and the area around the neck. The system relies on cameras and reference markings that are applied to soft silicone. The gadget can easily be placed around your neck and calibrates itself using the markings and cameras.

Once calibration is complete, the system can record muscle movements in the neck area and pass them on to an AI. The artificial intelligence analyzes the recordings based on their comparison data and uses them to reconstruct the words that the person wants to say. In the new system, the AI ​​can be linked to a voice model that can even be trained with the test subjects’ voices. In this way, a synthetic voice of the user is created, which then reproduces their speech even though they themselves do not actively speak a single sound.

In initial tests, the system is expected to work better than conventional techniques that rely on electromyography or electroencephalography. Those responsible also carried out tests with the prototype in particularly loud environments. In such areas, systems that rely on microphones to amplify voices would fail. However, there are also researchers who have found a solution for this.

The new AI system could enable patients who have temporarily or permanently lost their voice to speak again. And the technology could also prove useful in other situations. Namely when users need to communicate with each other in quiet environments. The AI ​​could then interpret the muscle movements and forward them as a text message or play them out to the other person via headphones.

However, it will probably still be a while before the system is ready for the market. The researchers want to improve the first prototype even further and implement support for multiple languages. They are also already thinking about how the device can be connected to other devices such as smartphones or tablets in order to integrate communication even more smoothly into the users’ everyday lives.

Robots are already being used here