Over the past few years, several projects have been launched in the world that create digital avatar technology using artificial intelligence.  These avatars look and can speak as real people and their main area of application is video content production and localization. In this post, we will tell you how AI avatars work based on the example of four companies developing them.

Projects dealing with creation of digital avatars usually offer a broad choice of ready-made characters. They can be used to produce educational videos, different video instructions, presentations, etc. 

Besides, these services usually envisage a possibility of generating a personal avatar of a real person with his/her appearance and voice based on the video.   So, for example, it is possible to create a digital version of company’s CEO to save time and money on production of video with his/her participation.  

Another advantage of digital avatars is their ability to speak many languages. That is why it is easy to create multilingual video sets with their help. 

British startup Synthesia was founded in 2017 . The service provides a choice of more than 160 ready-made avatars that can speak 130 languages. Also, a personal avatar can be created on the company’s website.

The Synthesia AI avatars are used to create educational, selling and marketing videos. Such large companies as Xerox, Zoom и Heineken are among the customers of the service.

In 2023, Synthesia received $90 m of investments (including the investments from American giant manufacturing chips Nvidia).  So, more than a year ago the company was appraised  at $1 bn.

The company does not disclose its sales and income indicators, though it declares  that it “constantly ensures three-digit growth”.  In particular, according to Synthesia, the number of service users in 2023 grew by  456% compared to the previous year. 

Since 2010, Canadian company Vidyard has been engaged in production of video content , focusing on creation of selling and marketing videos. In 2024, the service made it  possible to generate videos with AI avatars.

Vidyard’s focus area is creating personal ultra-realistic AI avatars . To that effect, a user should make a two minutes’ video using a web camera, based on which AI will learn. Vidyard also has a set of stock avatars, but it is significantly more limited[AL1]  than that of Synthesia.

A user may create a video with AI avatar based on the existing scenario setting it as a prompt.  Besides, one can use a scenario generator from Vidyard that will write a text for the user. The service accepts scenarios in 28 languages – the video can be voiced in them.

Also, Vidyard has a browser solution  for screen and web camera recording that can be used to create screencasts, among other things, with participation of AI avatar.

The company has a significant customer base. According to Vidyard, more than 12 million users and 160 thousand companies use its services. In April 2024, Vidyard received financing in the amount of $15 m  from Canadian government organization Export Development Canada (EDC).

 [AL1]Точного числа не приводят

Hungarian startup Colossyan in the beginning of this year raised  $22 m of investments. The main area of startup’s focus is creation of educational videos and trainings with AI avatars. 

 Colossyan has quite a big set of 150 stock avatars. Also, the service allows to create own AI avatars. Colossyan speaks 74 languages. There are several functions that mark the company out from competitors:

• An option of adding interactive elements to the video - for instance, quizzes. 
• Using up to 4 avatars in one video.
• Avatars in the video can be in three formats - entire body, bust shot or in the circle in one of screen corners with a presentation in the background.  
  

American company HeyGen offers a broad choice of ready-made AI avatars whose clothes can be changed. Besides, there is an option to generate own personal AI avatar, moreover, of different quality: Instant Avatar, Finetune and Studio Avatar. Depending on selected quality generation time and price change. Also “speaking photos” can be made on the website - that is creating animated avatars based on photos or pictures. 

AI avatar’s voice in HeyGen can be generated individually or selected from more than 300 ready-made versions, and avatars can speak more than 40 languages. The service envisages AI assistant based on ChatGPT to write scenarios to the video.  

In 2024, the startup is supposed to receive $60 m financing, thus the appraisal of HeyGen will be about $440 m. The company has demonstrated impressive financial growth having increased the revenue year on year from $1 m to $20 m only in a year .

American consulting company Gartner in its 2023 research makes a forecast  that marketing content share created by generative AI with human participation will grow from 2% in 2022 to 30% in 2025.

At the same time, the use of AI avatars in digital communications, according to the authors of the research, will have even more explosive growth: from the same 2% in 2022 to 70% in 2025. 

In Russia AI avatar technology is developed by Future Crew team.  For more details about project Altego Digital Avatar please read here.