artificial intelligence at the service of translation


Science fiction has imagined it for decades. Translation in real time Directly to the ear is now possible thanks to the latest advances in artificial intelligence, which allow a fluid dialogue without the language barrier.

At the Consumer Electronics fair in Las Vegas, CES, Several companies are promoting their increasingly sophisticated devices, allowing you to converse in multiple languages ​​with just a small device that fits your hand or wireless headphones that is barely noticeable.

Headphones "Pilot" by Waverly Labs (red, white and black with an elegant design) translate into 15 languages and cost between $ 180 and $ 250. If both sides have the device, you can speak directly.

During a brief demonstration in Las Vegas, an AFP journalist The French speaker was correctly translated by Andrew Ochoa, head of Waverly Labs, a company founded in 2014 in New York.

On the smartphone of Ochoa, the application also provided words in writing, in case it could be useful.


The handset, through the smartphone, "hears the voice, sends it to the" cloud ", transcribes, translates and synthesizes" vocally, explains Ochoa. As a result, it has a few seconds delay. "It's a" consecutive "translation and not exactly simultaneous, he adds.

The equipment and programs for translation and interpretation via speech synthesis have recently reached the market because the technology has not been sufficiently developed to date At least two years ago

"We are applying the latest capabilities of artificial intelligence, such as automatic learning or & Neural network & # 39; (computer system inspired by the connections of the human brain) to the models of translation ", he explains, also evoking the enormous progress of the vocal recognition.

Now "these technologies are mature enough to make it possible," says Ochoa, who says so has sold 35,000 pairs of headphones in less than a year, especially hotels, which obviously have to be able to communicate in many languages ​​with their customers.

Less discreet, the WT2 headphones from the Chinese company TimeKettle They work the same way and will be presented this week at CES.

You do not need a smartphone, however, to use Pocketalk, from the Japanese Sourcenext, which has in the viewfinder the Olympic Games in Tokyo in 2020because "many Japanese speak only Japanese," says Richard Gallagher, communications officer.

According to him, the small device, which looks like a cell phone of the ancients, can translate 74 languages for $ 299.

"Thanks to the automatic learning, you always understand better while using it" and "adapts to the pronunciation" of the user, he continues, while demonstrating.

"It's an excellent way to interact with artificial intelligence," he says, stating that they have already sold 200,000 units. Once again, it was the hoteliers who were very interested, but also traders and taxi drivers, among others.

the iFlytex porcelain, a maximum weight of artificial intelligence and speech recognition in Asia, is also present at CES with Translator 2.0, which translates Chinese into thirty languages ​​and vice versa. The latest model costs around $ 400.

L & # 39; Dutch Travis, which now manages 100 languages, claims to have already sold 120,000 pocket translators. They cost between 200 and 250 dollars.

The powerful Google, very present in the field of translation with his famous software Translate, inaugurated the dance in 2017 with the launch of "Pixel Buds", hearing aids with speech recognition.

At CES, The American giant presented a new translation function on Tuesday through his virtual voice assistant in many products, smartphones or connected speakers, which can now also listen to one language and translate it orally in another.

The next stage, according to Andrew Ochoa, of Waverly Labs, is the truly simultaneous translation, which technology does not allow yet.



Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.