TLDR: You can use LTX 2.3 to introduce audio into existing videos, with limits.Let me state right off the bat that I don't have the slightest idea what I'm doing.After studying a great many workflows, I managed to cobble together something that works, though I could not tell you how. For all I know, there may well be a more efficient or effective way of doing this, so if someone has a better method, I hope you'll upload it. And if someone wants to use this workflow as a template for creating something better, please be my guest.That out of the way, what does this workflow actually do? It is technically a V2V operation, but instead of adding more video to an existing video, it runs an existing video through the LTX 2.3 generating process, including the option for LoRas, adding (or replacing) audio at the cost of very slightly altering the original video.Why not just use MMAudio or Hunyuan Foley or some such? Frankly, they're probably better, but I figured why go to the complication of getting them set up when I already have LTX and its dependencies downloaded? A mere week and several hair-pulling sessions later and I totally don't regret that decision at all. So now I inflict my obsessive mess upon all of you with the full knowledge that it almost certainly won't be useful for the majority. But hey, more alternatives are never a bad thing.Now, the limitation. Given it wasn't designed for this purpose, LTX actually seems to do pretty well with syncing up sound effects and ambient noises, but lip-syncing is basically impossible unless you have a very clear closeup on a character's face and already know exactly what they're supposed to be saying. Either that, or it's a prompting skill issue on my end.Finally, be aware that, despite a loathing of subgraphs, I like my workflows very compact, so I apologize if it looks a bit overwhelming upon loading. I did try to arrange everything more or less sequentially and added notes to explain the adjustable parameters. However, this does assume some existing knowledge of how to run LTX 2.3, so maybe don't choose this as your first workflow if you haven't used LTX before. I also recommend making use of Comfy's link hiding function.For the two people who still don't use ComfyUI Manager, these are the required nodes:ComfyUI-GGUF: <a target="_blank" rel="ugc" href="https://github.com/city96/ComfyUI-GGUF">https://github.com/city96/ComfyUI-GGUF</a>rgthree-comfy: <a target="_blank" rel="ugc" href="https://github.com/rgthree/rgthree-comfy">https://github.com/rgthree/rgthree-comfy</a>ComfyUI-KJNodes: <a target="_blank" rel="ugc" href="https://github.com/kijai/ComfyUI-KJNodes">https://github.com/kijai/ComfyUI-KJNodes</a>ComfyUI-VideoHelperSuite: <a target="_blank" rel="ugc" href="https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite">https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite</a>And additionally for the WebP version:ComfyUI-WebpToPNGSequence: <a target="_blank" rel="ugc" href="https://github.com/Sekiun/ComfyUI-WebpToPNGSequence">https://github.com/Sekiun/ComfyUI-WebpToPNGSequence</a>My thanks to just about everyone who has uploaded an LTX 2.3 workflow in the past few weeks, it was all very informative and supremely helpful.A particular shoutout to @mylo1337 for their LTX 2 Foley workflow which inspired this idea.

LTX 2.3 Add Audio to Existing Video by WW

For those rare and foolish few who, like me, never changed Wan's default output in SwarmUI off of webp prior to that being changed. It is marginally less flexible than the video version, but it gets the job done.

Files with SHA256: 77f614d7959fe59202732ce8df9a20ad83b001da6a9a7b6128b8088d35049dce