Yeah, I admit, there's another Lily Loud voice model available. However, since my voice model is 300 epochs as supposed to the other one by LinsTLHfan having a much lower 80 epochs, I think mine should be better. Example uploaded here is directly from the source where I've made my model from, voicemy.ai, but the model is tested working on Kits Ai as well (vocals only for copyright reasons): Try the voice in Kits AI: HuggingFace Mirror:

Hey just a few questions what type of target rate did you use? 40k or 48k? Which pitch extraction did you use?

According to the post-created vocals it's seems to be 48kHz. I've made it via voicemy.ai and since the only thing I've done is upload the vocals and let the site create the model I have no idea what extraction method I've used at all. But I assume it's used whatever is compatible with Kits AI since the model uploaded there is the same one as the HuggingFace mirror. However, being a higher pitched voice, I recommend a +12 semitones at least for the voice to sound better, especially with deeper pitched voiced songs. (go whole octaves like +12 semitones or + 24, you can go in between, but it's not recommended for song covers).

B A B Y

This is so Perfect!!

It sounds so Quiet!