The international group consisting of the American and German researchers achieved progress in business of interpretation of the signals arising in a brain of the person at natural continuous informal conversation. And data of this interpretation were used for transformation of the “mental” speech to its text representation with enough fine precision. Further development of this technology can become the first step on the way of creation of systems of a mysleupravleniye computers and other equipment, new technologies of communication between people and many other.
The created Brain-to-Text system made record of the signals which are read out by means of a matrix of ECoG-electrodes (electrocorticographic) located in the corresponding places opposite to frontal and lateral areas of a cerebral cortex. Seven voluntary participants who were in one of medical institutions where they took a course of clinical treatment of the nervous diseases which are available for them including epilepsies were involved in experiments.
All participants read aloud and loudly standard text which consisted of a limited set of certain words. Collected data of brain signals were “fed” to the computer which by means of algorithms of computer studying and the analysis drew the most probable sequence of the signals corresponding to certain words. After that, by means of rather simple algorithm, data were transformed to a text format.
Unfortunately, the accuracy of work of such system for the present leaves much to be desired. In normal conditions quantity of mistakes when determining of the words said by the person was up to standard to 25 percent. And in case the patient owing to any reasons was incapable to say words absolutely correctly, changing some phonemes at a pronunciation, the quantity of mistakes increased to 50 percent.
But scientists are not going to stop on the reached results. They explain high percent of mistakes to that the analysis of data was made by means of the universal software, but not specialized which has to work much more effectively. When such highly specialized software is created, the percent of the mistakes made by system sharply will decrease and such systems will begin those “bridge” which will allow to communicate with the people incapable to talk owing to various reasons.