The "You can edit" part is a bit iffy still.
Yes, I have generated some pretty good songs (this is not an absolute statement, I mean "pretty good" compared to the staff UDIO generates normally) with UDIO, but it wasn't easy. I'd say that the average is 1 in 20.
However, I have had some success by uploading my own chord progressions. Just today I did another experiment and that didn't work out so well.
My first experiment took my progression and used it in the Chorus of the song. Today's test instead too my complete progression, inserted a good 2 seconds of complete silence, then made a song that's more or less in the neighborhood of what I was aiming for.
Actually, while this prevents a fully developed song from being published "as is", it could work if I download the stems and then just cut my progression from the beginning.
One thing UDIO does well (when you get lucky) is voices. Some are exceptional. I have tried the competition and they all sound heavily autotuned. Udio has used autotune only once in my song and it was an R&B song where it actually was used as an effect.
Unfortunately, the "editing" part is very much touch-an-go. For instance, there was no way for me to cut my progression out of the song it created.
In another case, I had a great R&B voice and a pretty decent song, but the chorus started with some kin of orchestra hit that was just readful. I tried multiple ways to eliminate it, but nothing worked (until I downloaded the stems and mixed it myself).
The other thing they need to work on is the lyrics regeneration.
Let's say you write a song and given its rhythm you realize that you need different lyrics in parts of it.
Theoretically it's possible (if clumsy), but in practice, 9 times out of 10 the new lyrics sound like a preacher speaking in tongues. IT'S NOT ENGLISH! Not at all.
For instance, this is one of the songs I created today. https://www.udio.com/songs/cWiZoyDhq9eSmN9s9oufX7
It starts with my chord progression (just bass and guitar because somehow Mixcraft lowers any MIDI down half a step, not a huge issue for drums, but a problem for the piano) , then silence, then the singer sings my lyrics quite well with a decent voice (I have had better, but also much worse).
The next step is to "Extend" the song by adding another 30 seconds and more lyrics.
This is the result: https://www.udio.com/songs/6d2ht5KpL4GJj3YA7ijqLo
I suggest you listen to it with the lyrics displaying, to get the full impact of the nonsense lyrics.
If somehow it doesn't work, I'll be happy to publish the song.
To me it's clear they need an actual music producer to help them design editing tools that are actually useful.
Things like cropping, equalizing the various instruments, changing the lyrics are at best clunky and at worst they produce absolute garbage.
Yes, I have generated some pretty good songs (this is not an absolute statement, I mean "pretty good" compared to the staff UDIO generates normally) with UDIO, but it wasn't easy. I'd say that the average is 1 in 20.
However, I have had some success by uploading my own chord progressions. Just today I did another experiment and that didn't work out so well.
My first experiment took my progression and used it in the Chorus of the song. Today's test instead too my complete progression, inserted a good 2 seconds of complete silence, then made a song that's more or less in the neighborhood of what I was aiming for.
Actually, while this prevents a fully developed song from being published "as is", it could work if I download the stems and then just cut my progression from the beginning.
One thing UDIO does well (when you get lucky) is voices. Some are exceptional. I have tried the competition and they all sound heavily autotuned. Udio has used autotune only once in my song and it was an R&B song where it actually was used as an effect.
Unfortunately, the "editing" part is very much touch-an-go. For instance, there was no way for me to cut my progression out of the song it created.
In another case, I had a great R&B voice and a pretty decent song, but the chorus started with some kin of orchestra hit that was just readful. I tried multiple ways to eliminate it, but nothing worked (until I downloaded the stems and mixed it myself).
The other thing they need to work on is the lyrics regeneration.
Let's say you write a song and given its rhythm you realize that you need different lyrics in parts of it.
Theoretically it's possible (if clumsy), but in practice, 9 times out of 10 the new lyrics sound like a preacher speaking in tongues. IT'S NOT ENGLISH! Not at all.
For instance, this is one of the songs I created today. https://www.udio.com/songs/cWiZoyDhq9eSmN9s9oufX7
It starts with my chord progression (just bass and guitar because somehow Mixcraft lowers any MIDI down half a step, not a huge issue for drums, but a problem for the piano) , then silence, then the singer sings my lyrics quite well with a decent voice (I have had better, but also much worse).
The next step is to "Extend" the song by adding another 30 seconds and more lyrics.
This is the result: https://www.udio.com/songs/6d2ht5KpL4GJj3YA7ijqLo
I suggest you listen to it with the lyrics displaying, to get the full impact of the nonsense lyrics.
If somehow it doesn't work, I'll be happy to publish the song.
To me it's clear they need an actual music producer to help them design editing tools that are actually useful.
Things like cropping, equalizing the various instruments, changing the lyrics are at best clunky and at worst they produce absolute garbage.
Statistics: Posted by conticreative — Sun Aug 11, 2024 12:25 am