The Pulse
HeyGen Pairs Voice Director With One-Photo Avatar Videos
HeyGen announced Voice Director and Avatar IV on May 20, 2025, combining prompt-based control of speech with animated videos made from a single image. Voice Director shapes tone, pacing and emotion, while Avatar IV syncs speech, facial move
AI.info Team · Written with AI ·
HeyGen’s Voice Director gives creators more say over an avatar’s vocal performance, while the system itself still adapts delivery and expression to the script. The company announced Voice Director and Avatar IV on May 20, 2025, pairing prompt-based control of speech with animated video generated from a single image. The launch page now carries an October 11, 2026, “Last Updated” stamp, but HeyGen’s product-update archive dates the announcement to May 20, 2025: this is not a new 2026 launch.
Voice Director puts delivery in the prompt
Voice Director lets users guide how an avatar says its lines through natural-language instructions. HeyGen’s examples include “emphasize this” and “add excitement”; prompts can shape tone, pacing and emotion to fit the message. The feature was live on HeyGen when announced, according to the company’s post.
Automation remains part of the design. HeyGen says its Panda Voice Engine adapts vocal delivery to the script’s context, while its Expressive Avatars match facial expressions to the voice’s tone. That puts user direction and automatic interpretation side by side: creators can request a performance, but HeyGen’s systems determine how the instruction carries through voice and face.
Voice Mirroring adds a personal performance
The announcement also describes Voice Mirroring, which uses a short recording of the creator speaking the script. HeyGen says the feature replicates characteristics such as rhythm, emotion, pacing and personality, rather than only generating the same words in a chosen voice. The stated aim is to avoid recording each take again while preserving more of a speaker’s delivery.
That approach gives users a different route from writing prompts: instead of describing a vocal style, they supply an example of it. The launch post presents Voice Mirroring alongside Voice Director, but does not give a separate rollout schedule or pricing for either feature. It also does not specify technical limits for the recordings.
Avatar IV turns a photo into a performance
Avatar IV takes a photo and script and generates an animated video, with voice synchronization, facial movement and gestures. HeyGen says gestures respond to the script’s context; the company pitched the feature for uses including product launches, tutorials, ads and social videos. Its original announcement said Avatar IV was rolling out, rather than stating that every user could access it immediately.
The single-image workflow distinguishes Avatar IV from a recording-based digital twin: the launch post describes starting with one photo, not filming a full performance. HeyGen says the system can animate people, pets and other characters. The announcement does not provide independent performance measurements or explain how it handles difficult scripts, so its claims about realism and gesture quality remain the company’s own.
Avatar IV later reached HeyGen’s API
HeyGen later extended Avatar IV beyond its web app. In an API announcement, the company said developers could generate videos from a photo and script inside their own products, with support for angled or profile images and lifelike or stylized characters. The API post lists self-serve access on Pro and Scale tiers, with custom rates for Enterprise.
The later API release shows how HeyGen carried the photo-to-video engine into developer workflows; it does not change the original launch date for Voice Director and Avatar IV. The May 2025 announcement’s defining combination remains user-directed vocal delivery and a generated avatar performance built from a still image and script.