***Note***

A recent update to this node may break it for some due to it changing where it was loading ffmpeg. Old version loaded from system python. New version loads from the comfy ui install python.

If you have errors, you may need to install ffmpeg for the comfy install. Some people may already have it properly set up but if you dont you will get an error of

module 'ffmpeg' has no attribute 'Error'

This can be fixed by installing ffmpeg for the portable python comfy is running on with this.

From your comfy ui python_embeded folder open a cmd window

uninstall

python.exe -m pip uninstall ffmpeg ffmpeg-python python-ffmpeg -y

reinstall

python.exe -m pip install ffmpeg-python

V3

Latentsync 1.5 model

This was not a clean install for me, updating the node did not change things.

I backed up my old latentsync folder in case i wanted to use the old one i dont know if they work in the same node.

Then i removed the node

Then i reinstalled it

Then i downloaded all models from hugginface and it worked.

These are the models, i dont know if auxiliary is needed, i installed it but you may want to try without it as i dont think its all needed.. but not sure.

https://huggingface.co/ByteDance/LatentSync-1.5/tree/main

Note there are 2 nodes in the manager for latentsync, get the correct one with the most downloads.

V2

- added video audio loading

- added original audio loading

- added image base for sync

- added live portrait extention for latentsync to LP Sync.

V1.1

-added .flak loading and a fix for a set node bug.

-Added other latent sync node

-Added face fix

V1 - Release

This workflow adds lip sync motions to a video from a source audio file.

This has an optional group to isolate vocals from music, or backgrounds.

Have fun.

Following nodes used

https://github.com/ShmuelRonen/ComfyUI-LatentSyncWrapper

Optional if you want to separate vocals from bg/music

https://github.com/abdozmantar/ComfyUI-DeepExtract