Nvidia-sponsored Synthesia reveals AI-generated “expressive avatars”

Synthesia has revealed a new range of artificial intelligence (AI) generated digital avatars which can portray human emotion based on text instruction. 

The London-based company has presented its “Expressive Avatars” to synergize real emotion in the virtual world with users able to create professional quality videos without cameras and microphones.

Nvidia-sponsored Synthesia is said to have half of the Fortune 100 companies on its client list with its services utilized by more than 55,000 businesses, overall. The firm specializes in creating digital avatars for corporate presentations and training purposes.

Actors read scripts before a green screen to train the AI system with one demonstration showing basic information being fed into the program.

The three lines of text were, “I am happy. I am sad. I am frustrated.” The AI-generated persona in the video then responded by reading the relevant text with the matching tone and emotion.

Since its formation in 2017, Synthesia has prospered to the extent it reached a valuation of around $1 billion last year, after a $90 million funding round. The soaring success has put the company on a “unicorn” platform, amongst other UK entities such as Accel, FirstMark Capital, GV, Kleiner Perkins, and MMC.

The world's first expressive AI avatars are (finally) here