I'm using some code I previously wrote for the company I work for. (MAJOR brand). They dont mind I reuse some of the code, so is fine. Tons of resources went into it, and there is not point in rewriting it. The idea was new so I have done few tests post on Youtube, nothing else. Could create something to guide users if there is interest. About the audio, I didnt really put much effort into yet, was just testing to see what kind of results I could get form the video models. Kinda put it over the video and it sounded ok so. I think with a bit of work it will be very difficult to tell is AI at all. I used the Elevenlabs API, they are also improving their models all the time. I think I used version 2 of the model, and there is a version 3 as a beta available. I feel like audio is a solved thing, it is the video that is not there yet, if you just pay a bit of attention to the hands you can rapidly know is AI, wont be the case in a few months. I think we are very close to the point, when you wont be able to set them apart. Happy to answer any other question you may have.The tech has come a long way in a short time but some notes and questions for those posting videos above:
1. Beside the VEO 3.1 system, are you using others to help create the video (Claude, chatgpt, specialized AI video tools, etc) or stand alone in this tool?
2. Where are you post these besides a Youtube page and maybe socials?
3. Some of the word tracks are obviously not a human and even some of the way they say things sticks out as AI. How do you clean it up?