Hey Gen
Hype about new translator AI: This is the start-up behind the viral videos
Watch the video: Suddenly a “language genius” – AI interpreter lets US director speak German and French.
This video is currently causing a lot of buzz on social media.
Tech influencer and director Jon Finger suddenly speaks different languages - without ever having learned them.
The clip was viewed more than 5.8 million times on “X”, formerly Twitter, alone in just a few days.
The online tool “HeyGen” uses artificial intelligence to do the translation and places it lip-synchronously and with your voice on the original video.
For this to work, the video must meet certain requirements and be at least 30 seconds long.
In the text accompanying his post, Finger asks other users for feedback on the quality of the translation.
Users are impressed by the technology, but note that the translations sound somewhat robotic.
But there are also critical voices: Experts such as Google employee Lutz Mache note that the technology can also be used improperly.
Artificial intelligence has been used repeatedly in the past to produce fake videos and put words in the mouths of well-known personalities and politicians.
Once again, artificial intelligence is causing a stir: it lets users speak in all possible languages. Behind it is HeyGen, a company from the USA that was hardly known before the viral hit
By Victoria Robertz
Speak French or Hindi without ever learning it? The AI from a US start-up now makes this possible. A video went viral on the online social network Many others have now tried the software and shared their results online. They are fluent in Italian, Spanish and German, although they speak another language in the original. It’s hard to see that there’s an AI behind it.
We developed the AI Joshua Xu and Wayne Liang, founder of the start-up HeyGen. Xu studied computer science and Liang studied human-computer interaction. They founded the company in California in 2020. “We are building an AI system that could significantly reduce the cost of content creation,” Xu writes on his Linkedin profile. It’s not just about translations, but also, for example, videos that do not require cameras or actors at all.
The tool that Jon Finger used to create the translation video is currently in an open test phase. A video must be at least 30 seconds long to be included HeyGen Labs can upload. The AI translates, clones the voice of the person speaking and synchronizes the image within a few minutes. Currently, ten languages can be entered and translated into eight, from Italian to Hindi.
“The Future of Video Creation”
But an AI that translates speech and images is only one element for the company. HeyGen already provides a type of video studio on its website. Over a hundred avatars can tell human-like stories in more than 40 languages. The text can either be entered as a script or uploaded as an audio file. “Welcome to the future of video creation,” says the avatar at the end of the explainer film. The company’s promise: cost-effective videos without complex editing. In their blog, for example, the founders explain how their software can be used to create marketing campaigns or onboard new employees. The cheapest subscription currently costs just under $50 per month.
At the beginning of this year, the US start-up achieved annual recurring revenue of $1 million within a few months, a “significant milestone” from the CEO’s perspective. However, he attributes the steep increase in sales not only to the number of users, but to the further development of the AI environment as a whole. With the translation videos, the company attracted greater attention for the first time. Jon Finger’s video alone was viewed 6.6 million times.
Even if some users note that the voices still sound a bit mechanical, it can be assumed that this AI – like many others recently – will improve at a rapid pace. Experts are already predicting that live translations could also be possible in the future.
Others warn about the downside of the whole thing: On the one hand, the willingness to learn languages could decrease. On the other hand, there is a warning against misuse of the tool. So-called deepfakes appear again and again in the context of AI, for example in the form of calls. Deepfakes use AI to fake faces or voices in videos, audios and photos. According to the Federal Office for Security and Information Technology (BSI), current research results indicate that this trend will continue and that manual detection of counterfeits will become increasingly difficult. The increased availability of tools like HeyGen and the low-threshold access to them will make it easier for laypeople to create high-quality counterfeits, according to the BSI. Heygen shows how easy it is.