<ul><li>Lipsync with Minimax H3 Ref2VA. </li><li>There is no transcript or lyrics needed in the prompt. The model is smart enough to lipsync, as long as you prompt it to. </li><li>This is the default ComfyUI workflow for Ref2VA, with: <ul><li>Kijai's LightX2V LoRA and Kijai's sage attention patch enabled</li><li>Math to calculate the length of the audio track to match the number of frames</li></ul></li><li>Due to request from community, I made a version with 2 speakers for a podcast. This is mainly a matter of prompting for the speakers to take turns and to have their lips closed when there is no speech in the audio. </li><li>What generally works is to give the official prompt guide (<a target="_blank" rel="ugc" href="https://github.com/MiniMax-AI/MiniMax-H3/blob/main/.agents/skills/h3-prompt-writing/references/ref-en.txt">https://github.com/MiniMax-AI/MiniMax-H3/blob/main/.agents/skills/h3-prompt-writing/references/ref-en.txt</a>) to a large language model and ask it to generate a prompt to do what you want. </li></ul>

Minimax H3 Lipsync

Removed MelBandRoFormer as podcast do not require audio stemming. To prevent the lips from moving when there is silence in the audio track is a matter of explicitly stating so in the prompt.

Files with SHA256: f607fc20eeb30c727ee4c9116fb1627d1398982d34f95c13866782c0f4899cc7