Have you seen any information on how to create the .npy files for custom voices, though? That's where I hit a wall. There must be some process that takes a .wav or similar audio file and creates the (relatively small) history prompt data.

Edit: looking at https://www.reddit.com/r/singularity/comments/12udgzh/bark_t... now.