Upgrade or remix a complete song using the new model (v1.5 or even higher)
under review
F
Frank Bruns
I've been doing some folk-metal/medieval-metal songs and found the new model (v1.5) to have a superior audio quality compared to v1. I've managed to revise some of my v1 songs by using the 2-minutes-remix feature. The revised versions sound way better now in terms of clarity.
Some of my older songs are more than five minutes long, and I would love to be able to run them through the new model in one go. I don't care if that meant considerably longer waiting times for the generations. The more time-consuming part was creating them in the first place. ;-)
SoulNRnB From The Soul Mixtape
Absolutely agree with this, it would also “future proof” songs we create today. Future versions of Udio could update older songs we like with better audio, different vocalists, different lyrics etc etc
Ian Braunstein
R
Rolf Joseph
Revising complete songs in one go would simplify many things.
- Change text at any point
- have the lyrics sung again in a different language
- Generate lyrics in several languages for duets
- Change the instructions for song generation within the lyrics
- Even an “extend” and “inpaint” would be possible, which takes over the sound of the song
etc.
If this is implemented in this way, you will have taken a big step forward.
Majik Stasia
Being able to remix a song up to ten minutes long would be awesome. most of my other songs on other ai song makers are longer than 6 minutes, and being able to remix them to improve their sound quality in udio would be awesome :) - I have been working on a few songs all year and no matter what I do, I cant extend it to be the same/similar, or prompt or anything. its frustrating because I love the song and want to finish it :)
I am happy for the wait time for Udio to create this - I have waited so far six month for this song I have been working on so far to get it perfect :)
Also to have the function to create a song and extend a song longer than 2 min / 32 seconds would be awesome :)
Jussimir Pasold
Yeah, would love to see some kind of "audio upscaling" feature so we can fine tune the song quality with a "creativity" slider so we can get choose how closer it should be to the original song
U
UdioAdam
under review
We hear you, we get why this would be cool, but it is genuinely more challenging than you might expect.
J T Brown
UdioAdam Thank you for this assessment. For me, it's incredibly helpful to be heard as well as to hear that it's more complex than we might imagine "behind the curtain"
MEC D
UdioAdam Would you guys consider then adding the abilities to regenerate/remix a highlighted area without extending? And Adding a strength bar with Regenerate at one end and Remix at the other?
U
UdioAdam
MEC D, remix doesn't extend, and already comes with a strength slider! :) (or did you mean something else here?)
C
CuNi
UdioAdam thank you. Even just saying "It's more challenging than you might expect" is a welcome update on this! If possible, many of us most likely also wouldn't mind spending more credits at it (every 32s adds 2 credits to the cost) simply because being able to remix a song fully would be a godsend! =)
Pasquale T.
I think that for those who already have many songs with the 1.0 generator it is a long-awaited feature, doing it by hand the result is never the same and sometimes you waste a lot of time to obtain a similar result with the remix.
Gainn
This would be really helpful. Even if it's a background process that takes a lot longer than normal gens.
You can do it by moving the remix section along so it just overlaps the previous and then stitching them together in an editor, but obv no good for publishing on udio.
F
Frank Bruns
Gainn I tried the process of moving the mixing sections along, but it is quite random whether the remixing steps fit together in terms of audio character. Often you can simply hear that the parts resulted from different remix attempts. Every so often it worked for me, though. I just would like to have it in one go, so the audio character stays consistent by default.
Gainn
Frank Bruns I've done it several times and not had issues other than slight volume variance. I use the seed from the initial V1 generation (32s or 130) for the remix and maybe that's helping keep it in line? I've also been lucky that the overlaps were never really more than 5 seconds.
There is a weird bug with the remix slider showing a different value to what it gives on occasion. I always give it a wiggle before I hit go to be sure it's on 0.1