HeyGen launched HeyGen Voice on October 9, and it arrived at #1 on the Artificial Analysis leaderboard, an independent ranking of AI voice models. The release matters less as a model announcement than as the final piece of an avatar-plus-voice stack that lets a founder produce branded video without ever stepping in front of a camera again.

The core model, Avatar V, builds a digital twin from a 15-second reference video, capturing motion, gestures, and expressiveness. HeyGen Voice is free inside the platform and API; a Professional Voice Clone add-on runs $99/month and requires 30 minutes to three hours of training speech. Identity and voice now come from the same vendor, which is a workflow fact more than a technical one.

The reason to pay attention sits in HeyGen’s own survey of more than 1,000 small business owners: 71.6% have recorded a video for their business and then decided not to post it. Shelvers are 1.5x more likely to believe they’ve lost business to a competitor. The gender split is sharper than the headline. Among women, 43.6% say the camera drains them and 76.6% shelve finished videos; among men, those figures are 28.4% and 69.6%.

That’s a quantified version of the thing every founder knows but doesn’t say out loud. The friction isn’t equipment. It’s the recording.

The survey also contains the real moat. Non-users predict avatars will look fake (50.7% cite it as their top fear) and rate expected audience reception at 3.06/5. Actual users rate reception at 3.49/5. The perception gap is where HeyGen gets to compound.

The timing is legible against the rest of the short-form ecosystem. YouTube Shorts’ new anti-reupload regime rewards original formats, and TikTok is now rewarding original creators with in-feed commerce. The platforms want more posting from more operators. HeyGen is selling the shortest path between a 15-second recording and that output.

Sources