The SilentSpeller project should make it possible to transcribe speech into text without having to pronounce the words. This is possible thanks to a sensor on the tongue.
A research project by the Georgia Institute of Technology a new technology developed.
With it a device can record text without it being sent by a Person must be pronounced.
It is much more likely that the person's tongue and mouth are measured when speaking silently.
- < li>
A new technology wants to change this.
This is possible thanks to a sensor on the tongue.
In order for a device to convert speech into text, you usually have to speak out loud.
You should only use it to form words, not to have to pronounce them out loud.
The technology of machine-transcribing spoken language has made enormous strides in recent years. It is especially helpful for people with disabilities to become more independent. However, up to now the words always had to be pronounced so that the computer could write them down. The SilentSpeller project aims to enable people to have conversations without speaking or using their hands to type, sign or gesticulate.
It is a research project by the Georgia Institute of Technology where Thad Starner is a professor. Starner is a wearable technology pioneer and played a leading technical role in the development of Google Glass. SilentSpeller is based on an existing product called SmartPalate, which is used by orthodontists to help people with speech impairments and speech disabilities.
124 sensors measure tongue movements
SilentSpeller is a communication system that enables people to send texts with a high-tech tooth holder to spell words without actually pronouncing them, such a demo video and scientific paper now available on the prototype. The device works by tracking the movement of the user's tongue. Researchers claim that the system recognizes letters with 97 percent accuracy and whole words with 93 percent accuracy.
According to the work, the device reads tongue movements and contacts with the palate through the holder's 124 sensors. The data is then sent to a PC or smartphone via a USB cable. Right now, that means users of the prototype have a wire dangling from their mouths, but researchers believe they could eventually develop a wireless version that fits completely in their mouth. The device needs a tooth impression for a perfectly fitting holder. This should make it possible to dictate silently, send e-mails or SMS without having to pronounce the text in a voice. Although researchers have been working on the project since January 2020, progress has been slower since the corona pandemic.
The product is not yet ready for the market, but could at some point be used to help people with movement disorders such as Parkinson's or other tremors. Starner also sees potential uses in really quiet places, like a library, or really noisy places where speech recognition is otherwise not possible.
If you subscribe to the Digital-Push, you will be informed about news and rumors from the world of Whatsapp, Snapchat, Instagram, Samsung, Apple and Co. You will also be the first to receive warnings about viruses, Trojans, phishing attacks and ransomware. There are also tricks to get more out of your digital devices.
How it works: Install the latest version of the 20-minute app. At the bottom right, tap on «Cockpit», then «Settings» and finally on «Push notifications». Under “Topics”, tap “Digital” – et voilà!
My 20 Minuten
As a member, you become part of the 20-Minuten community and benefit from great benefits and exclusive competitions every day!