I asked #Udio to make a #DepecheMode song and here's what happened!

Up until now, Suno.com has been the big thing in AI music generation, but now there's a new kid on the block called Udio.com. I gave it a whirl and was interested to see how it behaved when you put in a band name as a prompt. Suno won't let you do that, but Udio handles it in a smarter way by replacing the band name with suitable prompts. Since Depeche Mode is one of my favorite bands, I decided to experiment with that and was blown away and decided to share. https://www.andrea.net/temp/Udio_How_it_handles_band_names.jpg

Disclaimer: I mean absolutely no disrespect to Depeche Mode or DM fans. I've been a huge fan of the band since the 80's - have all their albums as vinyls and cds, seen them in concert many times etc. I simply wanted to share the results of what Udio can do because I find AI fascinating. It goes without saying that - nothing - can replace the original band and their talent. Not even AI.

EXTRA INFO:
Since I've had a lot of questions, here is some additional info:

a. The prompt I used was "Depeche Mode". Udio replaced it with the following 6 prompts for the parts needed to create this song:
1. Male vocalist, Electronic, Synthpop, Rock, Alternative rock, Dark, Rhythmic, Melodic, Atmospheric, Nocturnal, Alternative dance, Mechanical, Passionate, Cold, Longing, Sombre
2. Ditto
3. Ditto
4. Instrumental, Instrumental, Electronic, Synthpop, Rock, Alternative rock, Dark, Rhythmic, Melodic, Atmospheric, Nocturnal, Alternative dance, Mechanical, Passionate, Cold, Longing, Sombre
5. Instrumental, Instrumental, Instrumental, Electronic, Synthpop, Rock, Alternative rock, Dark, Rhythmic, Melodic, Atmospheric, Nocturnal, Alternative dance, Mechanical, Passionate, Cold, Longing, Sombre
6. Instrumental, Instrumental, Instrumental, Instrumental, Electronic, Synthpop, Rock, Alternative rock, Dark, Rhythmic, Melodic, Atmospheric, Nocturnal, Alternative dance

b. Lyrics were auto-generated by Udio. They're kind of corny and instantly recognizable as AI lyrics, however for the purpose of an experiment, it's kind of cool to document this moment in history before they become more sophisticated.

c. Whenever you generate music in Udio, it will spit out two samples. Here is the "twin" sample that it generated originally: https://www.andrea.net/ai_music/Sync%20in%20the%20Veneer.mp3
Although the song has a lot of potential, it is also has problems, so it was easy to choose the version in the video.

d. When Udio generates clips, they are 33 seconds long. So to end up with a complete song, you need to "extend" the songs by generating additional blocks of music. To get this end result, it took hours and hours of generating various versions because 99% of them didn't work well with the initial sample.

e. The final song ended up being compiled of 6 clips in total. You can see me playing samples of them here: https://www.youtube.com/watch?v=I8dDs7pihQc

f. The next day I tried to make more DM songs but failed miserably. At first I thought that Udio had made changes to the software to stop it creating songs that are clearly from copyrighted material. But actually some people started to comment that they were able to make DM songs. For instance this guy managed to make several DM songs: https://www.udio.com/users/ebgdMCDYYfy9dzAFQCEPsu

g. The day after that, I tried again and was finally able to make more DM songs. However this time instead of writing in "Depeche Mode" as the prompt, I used the prompts Udio had originally replaced it with (up in section a.)

h. It feels to me like Udio originally trained upon many known goth band vocalists (I heard Peter Murphy and Siouxsie Sioux several times for instance). And perhaps their system mixes the voices together in random amounts. And whenever you happen to get a "gahan voice", it simply has used a larger percentage of his voice over others? Dunno!

i. If you're interested in the copyright issues, you might find this Rolling Stone article interesting that talks about how both Suno and Udio have clearly used copyrighted music to train on:
https://www.rollingstone.com/music/music-features/udio-ai-music-chatgpt-suno-1235001675/

-------------------------------------------------------------------------------------------------------------------------------------------------

The images have been generated with #StableDiffusion + the photon model using Dave Gahan as a prompt, so the likeness is not perfect. I could have separately trained a Gahan LORA of course, but that would have been overkill for just one video. The images were inserted "as is", so I never bothered fixing flaws like double ears etc. In fact I think the little freaks of nature give the video a weird and wonderful vibe. And as with the lyrics - the imperfections of AI in it's current state are interesting to document. In a years time everything will be perfect and pristine, so it will be nice to have this result of the current times.

#aimusic #aimusicgenerator #suno #udio #udiomusic #aimusicevolution