Liri can handle so many jobs in multiple countries/regions because she rents her face to Hour One, a startup that uses people’s portraits to create AI voice-over characters, and then appears in marketing and sales organizations around the world. Educational video.It is part of a wave of companies Revolutionize the way digital content is produced. It’s right Manpower.
Liri personally does her waitress and bar work, but she doesn’t know what her digital clone is doing. “It’s absolutely weird to think that my face will appear in videos or advertisements of different companies,” she said.
One hour is not the only company Take deepfake technology as the mainstream, Use it to make a mashup of real footage and AI-generated video.Some also hire professional actors Add life to deeply forged characters. But Hour One does not require any special skills. You just need to be willing to give power to your face.
A so-called “role” pool is being established in the first hour. It says there are about 100 in its books so far, and it will add more every week. “We have a large number of people who aspire to be in these roles,” said Natalie Monbiot, the company’s director of strategy.
Anyone can apply to become a role. Just like a model agency, Hour One screens applicants and selects the people it wants on its books. Monbiot said that the company’s goal is a broad sample of characters to reflect the age, gender, and ethnic background of people in the real world. (Currently, approximately 80% of characters are under 50, 70% are women, and 25% are white.)
To create a character, Hour One uses a high-resolution 4K camera to shoot a person talking and making different facial expressions in front of a green screen. This is the artificial part of the performance. Insert the generated data into AI software, which works similarly to deepfake technology. Hour One can generate an unlimited amount of videos, and the person can say anything he wants to say in any language.
Hour One’s customers pay for the company to use its characters in promotional or commercial videos. They choose a face, upload the text they want to say, and then retrieve a video that looks like a real person sending a script to the camera. The fastest service uses text-to-speech software to generate synthesized speech that is synchronized with the character’s mouth movements and facial expressions. Hour One also provides high-quality services of audio recording by professional voice actors. These sounds are again suitable for the actions of the characters in the video. Hour One said it has more than 40 customers, including real estate, e-commerce, digital health and entertainment companies. One major client is Berlitz, an international language school that provides teacher-led video courses in dozens of languages.
According to Monbiot, Berlitz hopes to increase the number of videos it provides, but it is difficult to achieve it with real actors. They had to ask the production staff to make the same setting with the same actor over and over again, she said: “They found it was really unsustainable. We were talking about thousands of videos.”
Berlitz is now working with Hour One to generate hundreds of videos in minutes. “We are changing studios,” Monbiot said. “Humans don’t need to waste time shooting.”