Ok So ffmpeg is kind of magical. I have zero experience with audio and video editing really.
I want to take my old videos (like 100 of them) and swap out my voice with an AI voice.
I tried the elevenstudios.io dubbing feature where you submit a video and it does it automatically. It was horrible. Even when tried to use the studio feature to customize it it got worse.
Sooo...workflow is:
extract the audio from the video
ffmpeg -i "Vb6 Pcode For Loops Disasm.mp4" -vn -acodec copy ./forloops.mp4
split it into < 5 minute clips
ffmpeg -i forloops.mp4 -f segment -segment_time 299 -c copy parts/output%09d.mp4
reintegrate the clips back into the video
ffmpeg -i "Vb6 Pcode For Loops Disasm.mp4" -i final.mp3 -c:v copy -map 0:v:0? -map 1:a:0 forloops_new.mp4
Ill be damned if ffmpeg on the command line cant do it all! (except the ai submission)
I am mostly posting this here so I dont loose it. I will probably write a GUI to manage the cmdlines and paths shortly. Im amazed though..Still lots of voodoo so thanks to everyone whos cmdlines i ripped off from google searches!
Came out well, you can see the output here. Check the boundaries between the clips at the 5 minute marks.
I wish the voices were a little better, but for now its acceptable.
Also to the ffmpeg authors, you guys are fricking ninjas.
So I ended up writing a quick gui tool to help with the processing. AI voice conversion of a video is now reduced to two clicks. (select video, then full auto to convert)
I have spent ~$200 on AI credits on this project plus another $800 for the vids I initially had professionally edited. I had been putting this off for a year, so glad its finally done even though it took an entire holiday weekend.