Ramazan Eles 
Synthesia, an AI startup funded by Nvidia, unveils its new “Expressive Avatars” technology. This innovation allows AI-powered avatars to convey human emotions and movements, making them ideal for corporate presentations, marketing campaigns, and training programs.
Challenges of Existing AI Avatars Addressed
While AI for generating realistic videos has made significant progress, as seen with OpenAI‘s Sora video generator, challenges remain. These challenges include distorted body parts, unnatural backgrounds, and audio-visual desynchronization (lips not moving in sync with speech).
Synthesia’s latest update tackles these shortcomings. Their approach involves training the AI using real humans reading scripts in a studio environment. This process helps the AI capture accurate lip movements and deliver more natural emotional expressions.
Victor Ribarbelli, Synthesia’s CEO and co-founder, highlights a critical factor previously missing in AI avatars: “Unlike humans, avatars don’t understand what they’re saying.” This lack of comprehension resulted in a monotonous delivery, devoid of emotional connection.
In the studio, the AI was trained to respond to basic emotional prompts like “happy,” “sad,” and “frustrated.” This training allows the AI to generate the corresponding facial expressions and vocal tones for each emotion.
Multilingual and Customizable Avatars
These new “Expressive Avatars” come in over 130 languages and can provide closed captions. Additionally, Synthesia offers the option to clone a user’s voice for a truly personalized experience.
While Synthesia’s English language avatars deliver the most advanced and human-like experience, a recent test by Cointelegraph revealed a gap in quality for languages like French, German, and Spanish.
Synthesia boasts an impressive client list, including at least half of the Fortune 100 companies and over 55,000 businesses across various industries. Some notable clients include Zoom, Xerox, Microsoft, and Reuters.
A UK-Based Unicorn with a Strategic Focus
Founded in 2017, Synthesia is a UK-based company that has achieved a valuation close to $1 billion. With major backing from Nvidia, a leader in AI chip development, Synthesia has carved a niche for itself by focusing on creating business-oriented, human-like avatars.
By specializing in corporate-focused AI avatars, Synthesia avoids the intense competition witnessed in the general-purpose chatbot market, where players like OpenAI’s ChatGPT and Google’s Gemini vie for dominance.