Jump to content

spring song - my first AI-assisted song


equality

Recommended Posts

I had to test it :rolleyes:. I have used AI in photography before. An instrumental song (Little  Creek) with newly written lyrics was used as a template. Fast as hell, Suno.com delivered a number of suggestions where the song had been redone and different singers and some vocal harmonies were added. I chose the one I thought was best and imported the stems. I don't think the sound quality was particularly good and the vocal track lost some nice choirs along the way. I tried to clean up the vocal track but the pronunciation is sometimes really strange. The word "play" sounds like the French word "entré". "Neither" is sometimes pronounced "neether" and sometimes "nyther" (as in "eye"). Anyway, I am amazed and impressed! I can't hear that instruments have been added. However, Suno removed my piano. Added one of those, brushes, balalaika and organ. Is this the death for songwriters?  I would like to be able to just "bring home" a vocalist without changing a lot of the rest of the song. Maybe that's possible. I haven't explored all the possibilities.

 

Edited by equality
  • Like 4
Link to comment
Share on other sites

  • equality changed the title to spring song - my first AI-assisted song

Interesting. Vocal placement in the sound space is very... spacy in headphones, and it has a super-phasy and breathy sound. Nice little tune overall. Some timing weirdness with a couple pianos notes around 1:13.

I'll probably have to try this out with one of my own songs for fun sometime.

  • Like 1
Link to comment
Share on other sites

Mikael
I listened to your previous post (Little Creek -nicely done!) and then this one with the added in vocalist and other additions and gotta say I'm impressed with the result.

I tried Suno.com -- put in a phrase and it came back with a pretty good song including lyrics that made sense.  The vocalist sounded a lot like yours.

Not sure how I feel about all this but have to admit I'm impressed.

-Bjorn

  • Like 1
Link to comment
Share on other sites

9 hours ago, David Baay said:

Interesting. Vocal placement in the sound space is very... spacy in headphones, and it has a super-phasy and breathy sound. Nice little tune overall. Some timing weirdness with a couple pianos notes around 1:13.

I'll probably have to try this out with one of my own songs for fun sometime.

Thanks David for important feedback. As said, sound quality of the stems lag behind. I had to do some cleaning and eq:ing before accepting the results. The width is on me and I diminished it a bit. You're right, there is a mess around 1:13. I have a lousy ear for rhythm and beat. Hopefully, a  bit tidier now?

Link to comment
Share on other sites

7 hours ago, equality said:

Hopefully, a  bit tidier now?

Yes, better. And the vocal placement is quite a bit better. Previously, with heaphones, it sounded like it was floating in the middle of my head while the instruments were in a different space, more out front. I didn't recognize it as excessive width, but that must have been it. Nice work.

  • Like 1
Link to comment
Share on other sites

Hi

Very interesting - as you said the vocal can sound a little strange if one is being hypercritical, but believe me I've played with a couple of singers who sound much worse than this!!

Will AI be the death of music ?- only if we let it, I feel 

Nigel

 

  • Like 1
Link to comment
Share on other sites

On 3/14/2025 at 1:09 AM, bjornpdx said:

Mikael
I listened to your previous post (Little Creek -nicely done!) and then this one with the added in vocalist and other additions and gotta say I'm impressed with the result.

I tried Suno.com -- put in a phrase and it came back with a pretty good song including lyrics that made sense.  The vocalist sounded a lot like yours.

Not sure how I feel about all this but have to admit I'm impressed.

-Bjorn

Thanks a lot Bjorn! I wish Suno didn't change that much of the original concept of the song I uploaded. But we haven't seen the last versions of those algorithms, have we? Would be nice to hear your experience moreover. 

Link to comment
Share on other sites

18 hours ago, David Baay said:

Yes, better. And the vocal placement is quite a bit better. Previously, with heaphones, it sounded like it was floating in the middle of my head while the instruments were in a different space, more out front. I didn't recognize it as excessive width, but that must have been it. Nice work.

Thanks David for taking your time for a second look. I sort of get tunnel vision at home. That's why these forums came about, right? Glad you found an improvement!

Link to comment
Share on other sites

1 hour ago, Bajan Blue said:

Hi

Very interesting - as you said the vocal can sound a little strange if one is being hypercritical, but believe me I've played with a couple of singers who sound much worse than this!!

Will AI be the death of music ?- only if we let it, I feel 

Nigel

 

Right, AI is made by humans for humans and hopefully we may see it as a powerful tool, to get inspired of. I must say e g, I wouldn't come up with that balaika hook(?) if not triggered by the Suno-version. I did some serious attempts to clean the vocal file further but not lucky. Today I will try Ultimate vocal removal (UVR) as a last attempt to get rid of artefacts. It's interesting to see how what difference it makes to have the written lyrics in front of one's eyes. What she sings become almost crystal clear if you go along with what's written. Without it ,it sounds like a completely foreign language.😄 It reminds me of how much information severely hearing impaired people get from lip reading.  Thanks for listening an comments!

Link to comment
Share on other sites

On 3/12/2025 at 6:28 PM, equality said:

I tried to clean up the vocal track but the pronunciation is sometimes really strange

I like the female voice in your song but yeah, I have the same problem. Ive been using an online AI voice service for female backing vocals. A couple of the voices sound great tone-wise but pronunciation-wise, theyre hit and miss. They often turn my "ahhs" into "oohs". 

Im assuming this technology will inprove with time. It def needs to get better at listening more accurately.

  • Like 1
Link to comment
Share on other sites

24 minutes ago, T Boog said:

Im assuming this technology will inprove with time.

Most of u on this forum are more computer savvy than me. Im curious, in the near future, do yall see AI getting better at reproducing vocals more accurately? (With better pronunciation and lest weird artifacts)

  • Like 1
Link to comment
Share on other sites

Interesting experiment. AI is there to play with so why not. I have no idea what you did or what AI did,  but the vocals were what stood out to me really as hissy, like needing deessing. Maybe just my ears, the pollen is horrible at the moment. If I heard this in a movie I wouldnt think oh this is an AI generated song, and I do expect we will be hearing more and more AI generated music placements. I am trying AI video generating and its really fun. I havent tried music yet but probably sometime will. I like the song. Thanks for sharing it. 

  • Like 1
Link to comment
Share on other sites

20 hours ago, equality said:

Would be nice to hear your experience moreover. 

Mikael,  here's the AI song from Suno  that I mentioned above.   Not bad.  I was going for a spacy, ethereal kind of track . I like it esp the vocal.   Kind of wonder though where the AI engine or whatever it's called  gets the melodies and arrangements and lyrics.

I'm going to take the song down in a couple of days since it's not really mine.  

-Bjorn

 

Edited by bjornpdx
  • Like 2
Link to comment
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...