Generative artificial intelligence startup Hedra Inc. today announced it has raised $32 million for an AI platform that allows creators to generate lifelike and vivid videos of characters using text, audio and image inputs.

Andreessen Horowitz’s Infrastructure fund led the Series A round, with participation from the company’s previous investors, including a16z speedrun, Abstract and Index Ventures. The round brings the company’s total funding to $44 million since the last announced fundraise in 2024.

Hedra launched in 2024 to build high-speed, controllable video foundation models focused entirely on character-driven storytelling. At the time, generating lifelike characters, including voices and applying the capacity of motion and expression was a difficult process.

Since then, Hedra had also unveiled its first foundation model, Character-1. Now, the company has launched Character-3, which it calls an omnimodal foundation model for lifelike AI character video.

“We’re building the next generation of storytelling technology to empower content creators and enterprise marketers to tell narratives at scale on their own,” said founder and Chief Executive Michael Lingelbach. “Getting over the uncanny valley of compelling performance is the hardest frontier in video, and with our Character-3 foundation model, we’re devoted to crossing it.”

Since the company’s launch, Hedra’s technology has been behind the production of some interesting uses, including viral podcasts of AI-generated talking babies.

According to Hedra, its platform has gained more than 3 million users and powered more than 10 million videos. The company offers a web-based service called Hedra Studio, a content creation platform, which offers the Character-3 foundation model to take multimodal inputs, including text, audio and images and produce lifelike narrative-driven videos.

Users can start by inputting a script or uploading an audio clip. Or they can clone a voice using ElevenLabs Inc., Hedra’s integrated voice synthesis partner, to generate a custom voice in the accent, language or emotional tone of their choice. The character can be created using Hedra’s built-in image generator or uploaded directly.

Character-3 will then generate a full-body or upper-body animation that includes lip-sync, facial micro-expressions and body language. The rest of the scene, such as setting, backgrounds, camera angle and motion, can also be controlled by the user on the web interface.

According to the company, Hedra Studio provides a complete drag-and-drop command center interface for working with character animation, voiceover and environment controls. “This kind of creative agility is no longer just a nice-to-have — it’s essential for capturing attention in an increasingly crowded digital landscape,” said Lingelbach.

The company said it intends to invest heavily in pushing forward with AI model research for quality and controllable character video. The Hedra team has grown to 20 employees and intends to triple headcount by year’s end.

Images: Hedra