NEWS

EarCommand technology will allow you to silently give commands to gadgets, pronouncing them to yourself

EarCommand technology

Although it is quite convenient to control devices using voice commands, loud, clear phrases can disturb others. The EarCommand system, developed by scientists from the State University of New York, allows you to “read” the commands spoken by the user to himself, and this does not require reading brain activity.

Normally, voice commands can be heard by others, their recognition by gadgets can be interfered with by ambient noise, in addition, it can be difficult for devices to distinguish one voice from another, which will lead to errors in operation.

EarCommand technology, developed by a team of researchers from the State University of New York at Buffalo, is based on the fact that even speaking any words to yourself is accompanied by muscle activity and some bone displacement, as a result of which the ear canal is deformed in different ways. In other words, different “patterns” of deformations are characteristic of certain words.

The hardware component of EarCommand includes an earpiece-like device that broadcasts signals close to ultrasound into the ear canal. The signals reflected from the inner surface of the channel are received by the internal microphone of the earpiece, and the computer associated with the device analyzes these echoes, using a special algorithm in order to evaluate the deformation of the channel and determine which word is spoken.

In the experiments, users spoke 32 one-word commands and 25 commands consisting of whole sentences. At the word level, the share of errors was 10.2%, and at the sentence level – 12.3%. It is expected that as technology advances, the error rate will decrease. In addition, the equipment works even when the user is wearing a mask or is in a noisy environment and, unlike some systems that read spoken words, no camera is required to use the technology.

It is noteworthy that a similar EarHealth technology is being tested by colleagues from the same university for medical purposes – similar signals and echo readings from them are used to detect hearing problems, for example, diagnosing damaged eardrums or otitis media, which is a very common disease.