Synthesia unveils Express-3

A new avatar model promising sentiment-aware performances, improved lip sync and body motion, and twice the generation speed.

Synthesia has announced Express-3, which the company describes as its most advanced avatar model to date. The announcement, posted on Synthesia's official account, highlights three core upgrades over previous versions: avatars that can respond with sentiment-aware performances, improved lip sync paired with more natural body movement, and video generation that runs twice as fast as before. According to the company, Express-3 isn't limited to Synthesia's standard avatars. The model also underpins a feature called Style Avatars, which lets users generate stylized characters either from a text prompt or by choosing from a set of presets within the Avatar tool. The emphasis on sentiment-aware performance suggests Synthesia is pushing its avatars closer to conveying emotional nuance rather than just reciting scripted lines, a persistent challenge for AI-generated presenters. Faster generation times could also make the tool more practical for teams producing higher volumes of avatar-based video content, such as training materials or corporate communications, which has been Synthesia's core market. As with any vendor-issued product announcement, the specific gains in speed, lip sync accuracy, and motion quality come directly from Synthesia itself, without independent benchmarks confirming the improvements. Still, the update signals the company's continued investment in refining its avatar technology as competition in the AI video generation space intensifies.