Researchers at Cornell University have developed an EchoSpeech interface for recognizing quiet speech.
EchoSpeech glasses are equipped with a pair of microphones andspeakers that are smaller than the eraser on the end of a pencil and do not use a camera. The device sends and picks up acoustic waves that indicate changes in the dynamics of the facial muscles and mouth. A deep learning algorithm analyzes these echo profiles in real time with an accuracy of about 95%.
How EchoSpeech works. Image: Ruidong Zhang et al.
The received data is transmitted via Bluetooth toin real time on the smartphone, are processed and stored locally on the device. The developers report that EchoSpeech takes a few minutes to train for a particular user.
For people who cannot speak, this silent speech technology can be a great voice synthesizer. She can give patients their voice back.
Ruidong Zhang, co-author of the development
Most silent speech recognition technologiesare limited to a select set of predefined commands and require the user and the person they are talking to to look into or wear the camera. This significantly complicates the possibility of using such devices. In addition, a large data stream requires processing in the cloud, which violates user privacy.
In its current form, EchoSpeech canuse to communicate with others via your smartphone in places where speaking is inconvenient or inappropriate, such as in a noisy restaurant or a quiet library. The silent speech interface can also be paired with a stylus and design software such as CAD, virtually eliminating the need for a keyboard and mouse, the developers add.
Read more:
Scientists have figured out the nature of strange radio signals from a planet similar to Earth
A red halo flared up over Italy. Now its nature has been explained
Webb has found the oldest black hole in the universe