Liri can juggle so many jobs, in multiple countries, as she leased her face to Hour One, a startup that uses people’s likenesses to create AI-voice characters who then appear in marketing and educational videos for organizations around the world. He is part of a wave of companies rethink the way digital content is produced. And this has big implications for the human labor.
Liri does her job as a waitress and bar in person, but she doesn’t have a clue what her digital clones are up to. “It’s really a little weird to think that my face can appear in videos or advertisements for different companies,” she says.
Hour One is not the only company take mainstream deepfake technology, using it to produce mixtures of real footage and AI-generated videos. Some have used professional actors to add life to deepfaked personas. But Hour One does not require any special skills. You just need to be prepared to put the rights back in your face.
Hour One is in the process of building up a pool of what he calls “characters”. He says he has about 100 of them in his books so far, with more being added every week. “We have a queue of people dying to become these characters,” says Natalie Monbiot, head of corporate strategy.
Anyone can apply to become a character. Like a modeling agency, Hour One filters applicants and selects the ones she wants from her books. The company targets a large cross section of characters who reflect the age, gender, and racial origin of people in the real world, Monbiot says. (Currently, around 80% of its characters are under 50, 70% are female, and 25% are white.)
To create a character, Hour One uses a high-resolution 4K camera to film a person speaking and making different facial expressions in front of a green screen. And that’s it for the human part of the performance. By plugging the resulting data into AI software that works the same as deepfake technology, Hour One can generate an endless amount of footage of that person saying what they want, in any language.
Hour One customers pay the company to use their characters in promotional or commercial videos. They select a face, upload the text they want it to say, and grab a video of what looks like a real person delivering that script to a camera. The fastest service uses text-to-speech software to generate synthetic voices, which are synchronized with the mouth movements and facial expressions of the characters. Hour One also offers a premium service where audio is recorded by professional actors. These voices are again adapted to the character’s movements in the video. Hour One claims to have more than 40 customers, including real estate, e-commerce, digital health and entertainment companies. One of its main clients is Berlitz, an international language school that offers teacher-led video lessons in dozens of languages.
According to Monbiot, Berlitz wanted to increase the number of videos she offered but struggled to do so using real human actors. They had to have production teams creating the same setup with the same actor over and over, she says, “They found it really unbearable. We are talking about thousands of videos.
Berlitz now works with Hour One to generate hundreds of videos in minutes. “We are replacing the studio,” says Monbiot. “A human does not have to waste his time filming.”