A Brain-computer Interface is a process by which a computer connects to and stimulates the human brain. In the visual, auditory and tactile domains, sensory support interfaces are under development. German researchers are working on a special BCI that will be able to distinguish voices and enhance hearing aids.
The human brain evolved to be an expert at recognizing voices and sounds in a world full of vibrations. Ask any new mother, and she’ll tell you even a baby’s ability to recognize certain voices is far beyond that of any machine. Soon however, the Brain-computer Interface will compensate for dulled senses and enhance others beyond human potential.
Proof of Concept for the First Auditory Brain-computer Interface
Researchers at the University of Oldenburg in Germany are developing a Brain-computer Interface (BCI) that would be able to determine who exactly you are listening in a room full of other people conversing and noise. A paper on their BCI was published in Biomedical Physics and Engineering Express.New #German #BCI can single out voices in a room full of conversation.Click To Tweet
In the study, the 12 subjects listened to speeches by male and female subjects containing repetitive syllables and were asked to pay attention to only one of the two voices.
After training participants using electroencephalogram (EEG), researchers tested the system to see how accurately it classifies the EEG data to determine a voice. They found that the system first classified data with 60% accuracy, which is a good result but not suitable for practical use. Researchers altered the analysis time from 20 seconds to two minutes and the system’s accuracy rose to 80%.
Although there’s room for improvement, in the future, this technology could vastly improve hearing aids. While most Brain-computer Interface projects are focused on visual stimuli, an auditory-based system could also be of great help to paralyzed and visually-impaired patients, as the BCI’s sound recognition can also distinguish the clear source of particular stimuli.
“Merci Beaucoup” is Enough for the Brain to Recognize a Voice
In 99.9% of cases, two words of four syllables are enough to identify a familiar voice among a multitude of voices. This was the conclusion of a study by Julien Plante-Hébert, voice recognition doctoral student at the University of Montreal.
In the study, 44 people were asked to distinguish a familiar voice among ten men. Several voices had similar acoustic properties and varied lengths (between 1 and 18 syllables). The results showed that with words of four or more syllables like “merci beaucoup,” the rate of successful recognition was close to 100%.
Being able to differentiate between voices and surrounding noise, and to recognize a familiar voice, sounds like an easy, completely natural process. But this is the fruit of brain adaptation that happened in the course of a long evolution. Hearing capacities of the human brain are exceptional and, correctly identifying a voice in 92% of cases, surpasses all speech recognition machines thus far.
Memory is such a complex and multifaceted mental function that it is difficult to identify. Each type of information is conserved and recovered separately, and dozens of brain areas are involved, creating a complex network of interactions.
Just one of these facets is an auditory memory.
Hear that? It’s the sound of progress.
Just one more thing. As we often do, Edgy Labs wants to point out that new technologies purposed to recognize and single out any one person could potentially be used to infringe on that person’s privacy.
As we develop a set of ethics alongside our technologies, we must ensure that technologies meant to aid people don’t end up being used for the wrong reasons.