Here are some of the question people frequently asked me:

What did you use for your AI voice-over mods?

I used:

• WolvenKit for extracting the voice-lines from the game and creating the final mo
• ww2ogg for transforming the extracted game files into regular audio files
• RVC webUI for separating the vocals from effects and changing the voice of the voice-lines
• Wwise version 2019.2 for turning the audio files back to the game's format
• ffmpeg for joining the vocals with effects and other mass-manipulation of the audio
  

Consider using using my automation script!
Otherwise the basic steps are:


• Extract the voice lines you want to change from the game using WolvenKit.
• Convert the extracted .wem files to other format using ww2ogg (recommended) or WolvenKit's export tool.
• (Optional) Separate vocals from effects using RVC or UVR.
• Change the voice of the converted audio files using RVC (batch mode).
• (If you did step 3) Merge the vocals with the effects using ffmpeg.
• Convert the new voice lines using Wwise, Step 2 and further in this guide.
• Rename all the .wem files, remove the "_3F75BDB9" suffix. If your files have different suffix, try step 6 again. You can use PowerToys' PowerRename to batch rename the files.
• Put the renamed files back to your WolvenKit project folder where you took them and click Pack mod. Done!
  

There is also this fully-fledged Google Doc guide on vocoding voices.

I recommend https://weights.gg/! Otherwise there are many others on the internet, usually on HuggingFace. You can also train your own voice models in the RVC WebUI.

It took me a few days to figure out the process and get everything to work. Itself, the processing of the files took approximately 8 hours on my RTX 3060 and AMD Ryzen 5 2600 (12 threads).

It is not encrypted, update your WolvenKit, it's fixed in newer versions.

That issue arises when you convert files from .wem using WolvenKit, either wait for WolvenKit to fix the issue or use ww2ogg linked above to convert your .wem files.