Who has lost his voice per diseases such as a tumor or for other disorders could speak again thanks to the use of Artificial Intelligence.
Dysphonia affects almost one in three people
The dysphonia – this is the medical name to indicate the alteration of the voice – it is a rather widespread condition. Estimates speak of approximately 30% of the population being affected at least once in their lives.
The therapies available at the moment are very limitedwhether it is a temporary condition or whether you are faced with a permanent loss of voice.
Artificial Intelligence and voice: the novelty of the patch device
Now new research has developed a device that can be worn, applying it on the skin of the neck, in a non-invasive way. The patient almost does not perceive it.
This wearable deviceas the Anglo-Saxons call it, can identify the movements of the muscles of the larynx. In this way, thanks to the skills of machine learninghe’s able to reproduce the words that patients with dysphonia would have intended to pronounce. You can read the results in the scientific journal Nature Communications.
During the experiment, the researchers had eight adult patients wear this device. Everyone uttered five complex sentences. The results were excellent. The device managed to reproduce these words with an excellent 95% accuracy.
What is the machine learning?
Machine learning allows analysis a gigantic amount of data, facilitating the work of researchers. We know that Artificial Intelligence is now capable of learn from the information it receives, process it and decide practically independently.
The therapies available at the moment affect the quality of life of patients
The main problem of therapies to treat disorders affecting the voice is their duration, even when you have to face temporary inconveniences. Generally we turn to speech therapist or physiotherapist for the less important ones and others surgeon for others.
In any case they take a lot of time. It was calculated that it takes three to twelve months to go back to pronouncing the words exactly as you were used to doing before encountering the difficulty. Sometimes you are asked not to use your voice for a long time.
How does the new device work?
The new device would circumvent this problem. In fact, it could be used together with other treatments or during convalescence, so as to significantly improve the quality of life of patients.
The device consists of two parts:
and self-powered sensor, which reads the movements of the muscles of the larynx and transforms them into electrical signals, which thanks to a machine learning algorithm are translated into real words. A actuator capable of reproducing the sounds of recognized words.
The two components they are located inside a silicone patch of a few millimeters weighing seven grams, which is applied to the throat using biocompatible adhesive tapes. Basically a sort of plaster. The adhesive part can easily be changed every day.