Character 3 (https://www.hedra.com) is Hedra’s new video generation model (tested in June 2024): from an instruction describing a scene with a character, the definition of a text with voice and a reference image, an image and voice video is generated.

60 voices in 8 languages / accents (FR not available during the tests), recording possible.
Images (Flux, Recraft, Imagen, Sana, and Ideogram models) can be generated directly in the interface.

400 free credits per month (3-4 videos).

Texts generated by Mistral for audio testing.