Is there an AI service that is able to generate just voicing of the lyrics without any instrumentation? I tried to prompt the suno but it always creates a full song.
I guess for tomorrow it is time to try the stem splitting functionality of Ableton Live. It would be great to have an AI service that could sing the prompts without instrumentation per bbm whereas in Suno you have only ability to toggle the instrumentation only, or both.
2 Likes
not really shure, but you could try the manual mode in udio. should work somehow. they have a discord. they should know how to do it.
Iāve used this for stems (vocal/drums/instruments/bass)
@Heath thanks but That platform seems to just take stems out of songs. What Iām asking is saying āsing these verses on x bpmā with different styles and iād be able to get the one thatās working for my song or chop it. Suno is exactly there but itās just missing the ability to filter just the singing whereas you can filter the singing OUT so it should be really easy to implement.
1 Like
Gotcha, I see what you mean. Iām not sure of that. Sorry.
1 Like
There was a post recently with something that might be similar to this, but the lyrics were generated by the AI so Iām not sure how a human would interact with it. Iāll have to see if I can find the thread.
Ah, this is the one - I donāt know anything about it, but if the AI did this then itās likely it might be able to do what you want, I just donāt know how you would prompt it so youād have to look into what it actually does.
Check videos on Nobody & The Computer YouTube channel - I think you should find something there
*but if I remember it correctly, they were able to generate lyrics + instrumental and then used stem separator to get just the acapaella
2 Likes
Itās probably split using ableton live or such tools. I just created an idea on the ābarkā model github pages regarding this as I donāt see how it is possible to just create the singing parts but as you can filter them out it should be easy to just serve them as well. The underlying song does not really matter as long as itās tied to an bpm, it would serve as a great sampling material.
Also - for voiceovers that you could maybe cut in creative ways, try ElevenLabs - it wonāt sing, but might give you some cool vocals anyway - quality is really decent and you can create your own voices
You could use vocaloid studio, I mean youāll have to learn it like an instrument so you canāt just magic your way out of it, but it has the potential to sound quite good. Not sure if Iād call it convincingly natural or human, but it depends on your aim with the lyrics.
1 Like
Yeah I know there are some tools for these type of things. That one plugin thatās been around for years has one coco language to add the vocalisation. Namely, for the choir voicings iām after but Iād rather use the next gen free tools they are offering us if possible. Thanks for the recommendations anyways 
1 Like
2 Likes
Voiceswap.ai lets you replace your rough vocal with an AI voice modelled on a chosen singer. Iāve never used it though
1 Like