Speech-language pathology is an area of medical science based on the mechanics of voice production and the evaluation, treatment and prevention of communication. AI-assisted technology is now part of treatment options for conditions that affect speech, such as stuttering or the inability to control specific muscles after a stroke. 

UCLA bioengineers have created a device that translates larynx muscle movements into speech with incredible accuracy. This small, non-invasive device offers a promising alternative for those with voice disorders, providing an effective way to communicate during recovery.

Speech Pathology, AI & Wearable Devices

Everyone from healthcare professionals and medical researchers to students and graduates of institutions like the Ithaca College online SLP program can attest to the wonderful advances the ethical use of non-generative AI models has facilitated. 

AI’s unique ability to rapidly and efficiently analyze, compile, and produce results according to trends within the data analysis may come in handy with a unique magnetic phenomenon, magnetoelasticity. Magnetoelasticity describes the change of a material’s magnetic properties under strain. Using this concept and AI-assisted technology, a research team at UCLA led by Assistant Professor of Bioengineering Jun Chen has developed a thin, flexible device that adheres to the neck and translates the muscle movements of the larynx into audible speech.

The wearable device consists of biocompatible silicone and copper induction coils that generate electrical signals from muscle movements. When people talk, the movement of the vocal folds and throat muscles distorts the magnetic fields of the device, resulting in magnetoelasticity. When this happens, sensors in the device detect larynx muscle movements and produce electrical signals that an artificial intelligence model can read, interpret, and then produce output from. This output results in effective speech, allowing those with dysfunctional vocal cords to regain their voice function.

Tested on eight adults so far, it demonstrated nearly 95% accuracy in translating sentences. 

The research team plans to expand the device’s vocabulary using machine learning and test it on individuals with speech disorders. This non-invasive technology offers a promising alternative to current solutions and will be further tested and expanded to help those with speech disorders.

AI Applications Speech Therapy

In recent years, speech pathology technology has been developing rapidly. Automated speech recognition software and applications have been a highlight and some have been around for years. However, a huge advantage of AI models in speech pathology (as well as in general medicine) is the sheer volume of data they can draw from.

To work, AI has to be “trained” on input fed to it by the user. The AI can then store and remember all of this information and produce relevant data or output based on the data used to train it. Of course, humans are also capable of this, but it requires hours, perhaps even days, of sorting through test results, noting down the relevant data, and then comparing and checking it against itself. 

AI can be fed the data and produce the relevant stats, figures, or results in minutes. Also, since AI can be connected to audio equipment, it can recognize impairments and anomalies at much earlier stages than a human might be able to. There are even examples of some companies utilizing speech pathology AI that have products out to the market.

Finally, as it has been for the last few decades, AI can miraculously help develop and plan treatment for speech therapy clients. With its tremendous power of collecting, storing, remembering, recalling, sorting, and summarizing statistics and data, AI can look through patient records with unparalleled speed and efficiency and determine accurate and applicable treatment plans, considering the entirety of a patient’s history.