H3 Audio Sync Pixaroma

Makes a MiniMax H3 video sing the track you give it, instead of the meaningless sound the model would invent on its own. Put it between your latent and your sampler. H3 is unusual: it creates the picture and the sound at the same time, as one thing. That is why you cannot simply mute its audio and lay your own song on top afterwards, because the mouth was never moving to your song in the first place. This node drops your recording into the sound half and holds it still, so the only thing left for the model to decide is the picture that fits it. It works out how long the clip is by itself, so you never type a duration. If your track is shorter than the clip the node fills the rest with silence or loops it, whichever you chose, and says so on the node. If the clip is longer than about 15 seconds it warns you before the render rather than after, because H3 was only trained to about that length. This node only works with MiniMax H3. Find it by searching for h3, minimax, lipsync, sync, or music video.