Imagine being able to speak without moving your mouth. A team from the University of California, at Berkeley and San Francisco, found a way to convert brain signals into fluent, real-time speech. Using electrodes implanted on the brain’s surface, thoughts can be transformed into words without needing to move your mouth. A future where people can communicate telepathically using their thoughts may be closer than we think.

What Happens in the Brain When You Talk

Speaking may feel effortless, but a lot goes on inside the brain. It requires careful coordination between multiple systems that control your thoughts, muscles, and breathing. First, the prefrontal cortex figures out what you want to say. These thoughts then travel to the frontal lobe to a region called Broca’s area. Here, your thoughts are organized into words and sentences. Broca’s area is like the brain’s speech planner, making sure everything comes together grammatically and clearly. Meanwhile, in the temporal lobe, neurons in Wernicke’s area help ensure that what you are about to say makes sense.

Once your brain has figured out what to say, the sensorimotor cortex steps in to activate muscles in your vocal cords, mouth, and face to produce sound. As you speak, the auditory cortex listens to the sound of your own voice and sends feedback to Wernicke’s area, helping you to adjust as needed. This real-time loop is what allows you to speak smoothly and respond naturally.

However, for individuals with brain damage in the sensorimotor cortex, the brain may not be able to communicate with the muscles involved in speaking. This was the case for a 47-year-old woman named Ann. After experiencing a stroke at the age of 30, Ann lost the ability to walk and control her muscles, including those used for speaking. This study not only gave Ann a way to communicate but also enabled her to hear her own voice again.

How AI Restored a Woman’s Voice

The team began by recording her baseline brain activity. They implanted electrodes into the surface of her sensorimotor cortex. Ann was then asked to form sentences in her head using vocabulary from a bank of 1,024 words. Without making a sound, they encouraged her to “mime” or “mouth” these sentences so that they could record how neurons in this region normally activate.

These neural recordings were then fed into a custom-built speech-decoding system. The team designed it to recognize and match Ann’s unique brain signals with specific words. To make sure it sounded like her, they uploaded recordings of her voice from before the injury. Artificial intelligence helped fill in the gaps.

Now, Ann was ready to start talking. With the electrodes still implanted in her head, signals from Ann’s brain could be transmitted through the decoder and to a text-to-voice device. Just like that, her voice could be heard in near-real time, as if she were speaking normally.

Compared to other speech devices that sound robotic or have noticeable delays, this technology gave Ann her voice back. Even as her thoughts sped up, the speech decoder was still able to generate sentences accurately, without any interruptions. For people with brain injuries like Ann, this opens a new door of opportunity for unrestricted expression.

Conclusion

As this technology continues to develop, it has the potential to change how we all communicate. Instead of speaking out loud, we may one day be able to communicate directly through our thoughts. Rather than implanting electrodes into your brain, a wearable cap may be able to capture brain signals and transmit them to others. We are inching closer to a world where telepathy is no longer science fiction but a reality.