Hello audio friends,
for this last help - before starting to record - I would like to ask you something very very important for me.

Let me explain: For my song's ending I have written a simple two measures vocal melody without using words but just eleven times the particle 'na... na... na... na...' (so as to give the feeling of a kind of hypnotic tribal chant).

My dream would be for it to be sung by 'Difang Duana' ❤️ but - as you know - Difan is dead and only using a 'model' built with his voice (or other similar) and guided by my lead melody could I succeed to do.
The notes to be sung are only eleven and this, at least I hope, might make the operation easier... If it helps I can send you the sheet notes of the melody (it is in 12/8 triplets), although I don't think it serves the purpose.

Unfortunately, however, I have never created anything like this and I don't really know where to start for how to make such a 'voice model' but I have read that someone succeeds in this... and that is why I ask for your help (I’m on Mac).
I know it is difficult to realise this dream but ‘Never say never’, so the final question is: “How can I make Difang sing the notes I wrote for my melody?” There any guru who can guide me to create this wizardry?

〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰 〰〰

As a last resort the only alternatives I can think to Difang's voice are two:
>>> 1 - To be able to give that ethereal feeling, could be to use an african voice tribal chant (swahili, tanzanian, maasai, zulu style or something like that) Pitch Mapping it note by note to perform exactly the melody I wrote.
>>> 2 - Someone who owns ‘ACE Studio’ and runs my main melody MiDi file driving an african/tribal voice from the ACE Studio catalogue (whether there is something similar there).
What do you think, any different techniques?
I thank you and I await for advices/tips/ideas and your invaluable help...


> You can listen to Difang Duana a cappella voice here:

P.S.: I know that it is a very difficult task but if we succeed I might consider a small gift ($) to the person who will help me realise this little dream of mine.

 

Last edited: Nov 8, 2024

Hello ElMoreno, here is something to read and understand!

HOW TO EXTRACT VOCALS FROM A SONG - Unmixing a cake is not a piece of cake. But extracting vocals is.
https://en.audiofanzine.com/compute...to-extract-or-isolate-vocals-from-a-song.html

PITCHMAP - RE-WRITE MIXED MUSIC IN REAL-TIME
www.11dmedia.com/pitchmap.html

 

In the next few weeks I will start recording the song...
Has no one accepted the challenge yet?

 

I'd do an internet search... I bet you can probably find somebody that would be happy to sing what you want in the style and sound that you want for very little money. Bam, sorted.

 

Are you really convinced?
I wouldn't say it's really solved...

Difan is dead and, unfortunately, I don't see how anyone from the Masahi could be contacted for my African tribal voice... besides, do you really think he would do it for a very little money?
I doubt it, that's why I opted (and hoping) for the ‘Vocal Model’.

Anyway if you have a good internet ‘Prompt’ for specific research... give it to me, I'll gladly give it a try.

 

This is a technical niche skillset. I don't think a general call to randoms is going to give you a good chance of getting a serious answer. Think of all the garbage answers you get when you ask a simple question, like what is a HPF used for and now scale that up with how much most people (myself included) know about creating a working AI voice model.

Personally I would search this forum and others on AI voices, find people who have proven they already know what they are doing and either pay them if I could or if I can not, pester the shit out of them until they either tell me to f*** off into a live volcano or offer some advice that I find useful in completing my task.

good luck


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

You can search places like gigheaven (punto) com but honestly your floor for expenditures is going to be around 200 euro for a serious performer. Add an other zero if you want someone actually in Europe or the UK.

If you are looking for performers in the far east then it's a little outside my wheelhouse but there is a guy around here sometimes who has an avatar in a shark costume that is on the west coast and might have connections to the local scene which might lead to the faraway scene.

hustle on hustler

 

Last edited: Nov 9, 2024

I extracted Difang's voice and uploaded it to you as an MP3. Simply download it from the link and do a little cutting and editing.

https://mega.nz/file/Tn43gS6S#Yv4kPeMegd6bpz32OygK1Vyjw9LmqM2rUJ--OPrGjNo

Difang & Igay singing "Weeding and Paddyfield Song No. 1"

 

Last edited: Nov 9, 2024

Maybe I'm far out, because I've never used it, but isn't this exacrly what the free Rubin AI Replay is for?

 

More or less but, unfortunately, I still don't quite understand how control Rubin AI Replay to do what I want to do for my song...

 

Last edited: Nov 14, 2024

I tried to contact someone with a private message but only received negative replies.
If you happen to know someone who is available let me know, I would love to talk to them about how to realise the voice model I need.

 

Last edited: Nov 11, 2024

Check out that DJ kid from down south somewhere @curtified - he had a Discord and some other stuff about AI models and seemed internet kind if you came at him the right way. That's off the top of my mind.

I will add that there was an other guy on here who was making a model of Marie Fredriksson's voice for some Roxette unfinished songsn, I think it was on this forum but things get a little jumbled up in my brain. Search for it maybe it turns up.

If you already tried Curtified and he blew you off then keep on keeping on.

That's all I got without actually getting Huckelberried into the process.

Cheers.

 

yeah hit me up on a PM i can help however you need. Or you can make your own model on our website http://weights.gg and run it there or use our free desktop app http://tryreplay.io to drag and drop between your daw of choice.