Synthesia helps a TechCrunch reporter make an interactive avatar

In short: A TechCrunch reporter worked with Synthesia to create an interactive digital avatar that can talk with people, but only about a specific article.

What happened

TechCrunch reporter Dominic-Madori Davis said she created an interactive “digital twin” of herself using Synthesia, a company that makes video avatars for businesses. An avatar is a video character that looks and sounds like a real person, like a moving customer service kiosk on a screen.

Synthesia recorded a short sample of her voice and took photos in a small studio. The company then produced multiple versions, including “personal” avatars that read a script, and “interactive” avatars that can listen and respond.

Davis said the interactive version was trained on one topic, her reporting on venture fraud. “Trained” here means it was fed a specific set of material so it knows what to say. The result is “deterministic,” meaning it stays inside those boundaries, like a museum guide that only talks about one exhibit and redirects every other question.

Synthesia’s system combines several parts to make this work. One part turns spoken words into text, another decides on an answer, then another turns that answer back into a voice, and a video model animates the face as it speaks.

Why it matters

Tools like this could change how companies and public figures handle questions online, and they raise basic trust issues. Davis said the experience left her with mixed feelings, especially about whether people would accept avatars in journalism, where credibility matters. Even when an avatar is limited to a narrow topic, it can still feel unsettling, and wider use could increase confusion about what is real and who is actually speaking.

Source: TechCrunch AI