Nvidia-backed AI startup releases avatars that express human emotion
Synthesia, an artificial intelligence (AI) startup supported by Nvidia, has introduced a new upgrade enabling AI avatars to convey human emotions and movements.
On April 25, the company revealed its “Expressive Avatars,” which aim to express emotion based on text instruction for corporate presentations, marketing and training purposes.
This is it. This changes everything.
For the first time in history, #AIavatars are able to understand what they’re saying.
Powered by our new AI avatar model, EXPRESS-1, they can detect sentiment of a script and perform all the subtle nuances of human communication.
The… pic.twitter.com/uj3WIjcGlm
— Synthesia (@synthesiaIO) April 25, 2024
While generative AI has often been praised for its ability to create realistic moving images, such is the case with OpenAI’s Sora video generator.
However, AI is not without its flaws, especially when portraying humans, who are often shown with distorted hands or limbs, collaged backgrounds, or lips out of synch with speech.
Synthesia aims to correct this in its latest version, which was developed on actual humans reading scripts in their studio. This was done to help bots capture lip tracking and be more accurate in their emotive expressions.
Victor Ribarbelli, the CEO and co-founder of Synthesia, said in a video that the missing piece has been that, unlike humans, “avatars don’t understand what they’re saying” which previously led to the lack of facial response to emotions.
In the studio, they were trained to respond to simple prompts like, “I am happy. I am sad. I am frustrated” by conveying the correct facial expressions and tone associated with the given emotion.
The new avatars are also available in more than 130 languages, can provide their own closed captions and even clone users’ own voices.
Related: Sam Altman pushes ChatGPT mass adoption among Fortune 500 companies: Report
However, among the example models of avatars on Synthesia’s website speaking in languages other than English such as French, German and Spanish, the English language model is the most advanced and human-like, according to a Cointelegraph test of the model.
The startup reportedly has at least half of the Fortune 100 companies listed as clients and provides services to over 55,000 enterprises. This includes leaders in various industries such as Zoom, Xerox, Microsoft and Reuters, among others.
Synthesia is a United Kingdom-based company founded in 2017. On the heels of the AI boom over the last year, it has reached a valuation of nearly $1 billion with major backers like Nvidia — who currently dominate AI semiconductor chip development.
Due to its more narrowed approach —creating human-like avatars for business use — Synthesia has been side-stepping some of the hype and fierce competition seen between competing chatbot models like OpenAI’s ChatGPT and Google’s Gemini chatbot.
Magazine: How to get better crypto predictions from ChatGPT, Humane AI pin slammed: AI Eye
