Robot and robotechnology news
01.01.1970 //The thoughts of people will read on nervous impulsam.
American researchers from the university of Carnegie- mellon (Pittsburgh, the state of Pennsylvania) develop new firmware complex for the translation of speech from one language into another in the regime of real time. The special feature of system consists in the fact that she can record the phrases, which even not were pronounced aloud. As reports New Scientist Tech, the work of complex is based on the use of special electrodes, which record the electric pulses, sent to the muscles of face and language. These pulses then are converted into the text, which after transfer into the necessary language is reproduced by the synthesized voice. Thus, in order to say phrase, for example, in the Spanish, to user it is sufficiently simple to speak it about itself in its native language. Scientists, who conduct studies under the management Tanya Schulz note that the new machine translation system will be deprived of the deficiencies, inherent in the majority of analogous complexes. For users phrases it is not necessary to indicate aloud which theoretically must simplify the process of the contact between the foreigners. In the past year the specialists demonstrated the prototype of the machine translation system, capable of recognizing on the electrical signals of approximately hundred words in the Chinese Language and of converting them into the English and Spanish versions. The new modification of complex will be able to recognize the practically unlimited quantity of words. The secret of system consists in the fact that it records the phonemes, on the basis of which then are restored words themselves. However, until the work on the complex is distant from the completion. The accuracy of the recognition of unknown words is about 60 percent, what it is insufficient for the qualitative transfer.