In recent years, steady progress has been achieved in the field of brain-computer interfaces. Devices have been tested and utilized by paralyzed patients to reliably move robotic arms or computer cursors with great dependability.

Although such technologies can substantially enhance disabled individuals' freedom, at the moment there are no reliable methods of communication for those who are unable to talk.

However, researchers from the Netherlands and Germany have made great strides in that area as they developed a system that can instantly convert thoughts in the brain into speech with the clever use of machine learning.

Image Credit: ra2 studio via Shutterstock / HDR tune by Universal-Sci

The scientists were able to get an epilepsy patient with depth electrodes in her brain to utter words she only thought of in her head. The team published their findings in the highly regarded scientific journal Communications Biology.

Dr. Pieter Kubben from the Dutch Maastricht University Medical Center explained in an interview that a wide range of tests was performed on epilepsy patients that were monitored at the Kempenhaeghe expertise center for epilepsy.

These patients already had depth electrodes in their brains and agreed to partake in this study while on observation for epilepsy research. Thus, they were hitting two birds with one stone, so to speak.

Converting brain activity into speech

One particular series of tests focused on speech. First, a patient had to read aloud a selection of texts, after which a computer distilled the relationship between spoken words and brain activity using machine learning.

Finally, the system could generate audio directly from measured brain activity, even when the patient only imagined the words without speaking them out loud.

Christian Herff, the study's lead author, stated that the data models derived from brain activity from ordinary speech also work for imagined speech.

This extraordinary breakthrough is a significant step forward in the conception of a speech neuroprosthesis. It would allow those with severe speech problems caused by brain injury to participate in the program to speak again.

The current feasibility research was based on 100 different words. The researchers are now increasing their experiments with the aim of getting patients to convert entire sentences made up in the brain to speech.

For a more detailed overview of the study, be sure to check out the paper listed below.

Sources and further reading: