Nvidia-backed firm rolls out AI avatars that emotionally respond to text: All you need to know

tech.hindustantimes.com:

Synthesia, an AI firm supported by Nvidia, has introduced "Expressive Avatars," AI-generated digital characters capable of conveying human emotions through text inputs. This innovation is set to revolutionize the world of video production by reducing the need for cameras, microphones, actors, and lengthy editing processes.

Based in London, Synthesia's studio employs actors to read scripts against a green screen, teaching the AI system to replicate human expressions and emotions. The "Expressive Avatars" were demonstrated by inputting simple text lines like "I am happy," "I am sad," and "I am frustrated," with the AI-generated actor responding in the respective emotional tones.

Also read: Gigabyte launches AI powered QD-OLED gaming monitor in India: Check price, features and more

Synthesia's technology has gained traction among over 55,000 businesses, including half of the Fortune 100 companies, using digital avatars for presentations and training videos. Established in 2017, the firm secured $90 million in funding last year at a valuation of approximately $1 billion, cementing its status as one of Britain's prominent AI unicorns. Notable investors include Accel, Kleiner Perkins, GV, FirstMark Capital, and MMC.

Amid concerns about potential misuse, Synthesia has taken measures to ensure responsible use of its technology. Publishers must register as enterprise customers, and all content is subject to moderation. The company employs a "Know Your Customer" process akin to banking industry standards, preventing misuse by malicious entities.

Also read: OpenAI former executive shares why ‘human employees' will always be needed in the age of AI

Synthesia is actively preparing for global elections, implementing controls to thwart attempts by hostile actors to manipulate outcomes. Additionally, the company collaborates with the Coalition for Content Provenance and Authenticity, an AI consortium focused on implementing content credentials and digital "watermarking" to distinguish