This equal to customize vocal model, if this feature is real, I will stick to Udio forever, I makes mandarin songs with Udio but vocal is very repetitive, only vocal customization or vocal reuse (reuse the vocals from upload audio)

I am in the same boat, would love this feature!

I would really like this feature as well. Along with the ability to select the AI vocalist by tag and being able to select more than one for duets or outlining a full band that sings together.

This would be huge!

For myself the copy seed seems to be 90% voice. I am using 2:10 model 1.5. If I use seed '2' and put in the prompt, grunge, rock. 9 out of 10 times I get the exact same voice.

Does anyone else notice this?

In addition, a straightforward way to get it to work with voice cloning from an uploaded voice sample would be amazing. I know this feature work. But as i reported on Reddit, I can't get it to work--and it's not unique to me. Even YouTuber Bob Doyle couldn't get it to work (it only extends into a track without instruments).

This is especially desirable to me since the 1.5 model came out. I can no longer get a voice I like at all for my favorite music type using 1.5. Most of the vocalists I get when I try to create a power metal song are growly screamers whose vocal is mixed too far in the background. I just can't get a smooth powerful singer at all in 1.5. This isn't a huge deal as long as we still have access to version 1, but if at some point we can't use the original anymore I just wouldn't be able to make as good of song.

This can be done also using the upload feature. Generate a song with the voice you want. Download the vocal stem. Then generate a new song only instrumentals and download it, could be different genre. Overlay the vocal stem onto the new songs. Reupload this song and extend. This will use the original vocals as context for the new song

This would be so fun to have! I wanna make up a discography for the rock star character I created hahaha.