It's almost routine. When Ukrainian President Volodymyr Zelenskyy speaks into the camera of his smartphone, Eugen von Rubinberg knows he has to deliver. Because just minutes later, its algorithm takes the video from Ukraine and translates it into dozens of languages.
Eugen von Rubinberg is co-founder of the Swiss start-up Vidby, which uses artificial intelligence to automatically translate videos into more than 70 languages - almost in real time. "Our mission is to make video content accessible to anyone, anywhere in the world," he says.
As a result, Vidby superimposes a narrator's voice over the video, just as one would expect from an interpreter. It still sounds a bit like a computer-generated voice, but in the future Vidby will also recognize emotions and voice colors that result from the volume, power, resonance and tone of the voice and interpret them accordingly.
Technology from Switzerland has become an important tool for Ukraine. "Nowadays, the speed of information dissemination is crucial," says Iryna Borovets, Director General of Public Diplomacy and Communications at the Ministry of Foreign Affairs of Ukraine. “It's almost impossible to translate a video into 10 to 30 languages in a day using traditional methods.” Borovets says Vidby, on the other hand, can do it.
The amount of videos can hardly be managed otherwise. According to Ukraine's Foreign Ministry, more than 70 speeches by the President have been translated using Vidby, which equates to over 650 minutes of source material. Translated into up to 30 languages, this results in more than 10,000 minutes of new video material, which will be published on the websites of embassies abroad and on social media.
It is no coincidence that Vidby is providing its services to the Ukrainian authorities free of charge. A large part of the company's developers are Ukrainians and still work in the country today. "This is our small patriotic contribution that we are making," says von Rubinberg.
The company, which von Rubinberg founded together with Alexander Konovalov last autumn, claims that its translations are 99 percent accurate. However, the transcript must be prepared for this.
Words that are not to be translated must be marked beforehand. A fully automatic translation achieves an accuracy of 80 percent. Enough to understand the content. But not enough to look really professional.
For Vidby, Konovalov and von Rubinberg didn't have to start from scratch. In 2017, they already caused a stir with Drotr, a messenger capable of translating more than 100 languages simultaneously. The conversations in video conferences could be translated into more than 40 languages. But there was no money to be made from it.
That will change. Rubinberg puts the market for video and audio transcriptions at 30 billion dollars, a large part of which is in the USA. With Vidby, the start-up is now addressing companies, the media, non-governmental organizations, authorities and influencers. Customers include Samsung, Siemens, Generali, Cisco and Karcher. The latest addition is Harvard University.
"Everything on shares" is the daily stock exchange shot from the WELT business editorial team. Every morning from 7 a.m. with our financial journalists. For stock market experts and beginners. Subscribe to the podcast on Spotify, Apple Podcast, Amazon Music and Deezer. Or directly via RSS feed.