@@trueandobjective, yes, you must first remove the singer, then record a piano above, try audacity (or other software remover) for that (remove vocal) (automacatilly i dont know if its possible, i guess you cant)t
Dude, I just tried this. Beckys vocals pitched up 12 on my song.. It came out EXACTLY how I envisioned the song in my head. It went from blah to a fcking hit. This is powerful. I gotta buy the monthly plan. Im telling you, if you know how to use her, becky is a superstar. I feel like I have a young female prodigy that nobody has heard of yet lol
@@soundlearn i ran into the same problem u had. I was making songs that are out of my vocal range and should be sang by a girl. And you're right, your vocals still have to be decent. You can't upload trash vocals and expect good output. She sounds way better than me, but she's not a miracle worker 😂 you still have to be a decent singer for this to work.
Eventually, I would imagine we could design unique sounding vocalists just by talking to the AI like a producer talks to a human vocalist. Describing how we want them to sing and using trial and error to refine their performance to fit the song perfectly.
Question for you then. If I want to record a song vocals only no instruments and then get ai to add style (arena country for instance) is there an ai that will clean up everything and make a demo from my recording?
OK is it me or did the quality of this video jump up! Nice work. This is such a dope tool to use and kinda freaky how quick AI is learning to do stuff like this.
Similar to what Spitfire Audio do with orchestral players, they pay and ongoing royalty to the players. In the end this stuff is pretty much unstoppable
Ive used a similar platform, with other singers. It’s truly insane! So useful for producers who are either working solo on an idea or needs to pitch a tune as you said in the opening. Evej if just using tidbits in electronic music, it can be so great instead of sampling from accapellas and mevervreally gettingvwhat you want, having to meet with a vocalist, and by the time you meet you dont feel the track anymore 😂
Problem is probably in the terms and conditions - write lyrics and your song and upload and where or what is it used for to perfect the AI models ... AND ... maybe it ends up in a generative AI version somewhere, you'll never know... no? Did you read the T&Cs?
I feel like this is useful for proof of concept but feels soulless as something you’d release as a song. I think a useful tool utilizing this would be something like: 1. You could take your vocals, do something like this where it sounds more melodic and pitched correctly to give you something to hear to improve your own vocals after the fact. 2. Produce back up / crowd vocals to add layers to your production.
Yes, it is a 1:1 conversion in terms of performance. A bad, pitchy vocal with bad timing will still be a bad pitch vocal with bad timing, just with a different voice.
I think this is a great tool for zeroing in on your production. It seems much like composers use orchestra sample libraries to mock up soundtracks for movies. Once the director has approved the music then they go into the studio with a real orchestra. Also, for songwriters who are just trying to produce their own demos what a dream being able to audition different vocal models for a given song. You can't really experiment like that in the studio with real vocalists (unless maybe you are Steely Dan).
Whenever someone says "I don't think AI will be doing soon"........ my left eye kinda crinkles and the lips purse on the same side. One thing that is true..... is an AI will never be able to perform a flesh and blood live performance .... it won't be long before that's all we have :)
Hey thanks for the tutorial. You still need to sing pretty good to make this work, but it's very promising. Maybe do a follow-up video where you sing with different emotions and vocal strengths, i.e. from whisper to belting. I'm curious how the Audimee singers handle different dynamics. The problem if you want to do a Beyoncé type of track is that you still have to perform it close to her style and her vocal abilities to be able to translate it to a good (demo) track. The vocal synths (SynthV) are on the rise also, and I wonder if a hybrid of those would be a powerful solution. (Programming a synth voice is a completely different workflow, but maybe blending/comping that with your own voice and then feeding into Audimee will give a great result.) Another thing to showcase is the feature to create (train) your own voice model in Audimee, that could be great for songwriters/ producers to maybe lounge a semi-virtual artist.
I do have some information about what you wrote. It may depend on specific platforms, as their programming/features of AI rendered vocals might be slightly different - but I was able to basically whisper vocals in the correct pitch and the correct timing. When the audio track I uploaded was rendered by a female AI vocalist, it worked fine. The high notes were rendered as if she had to use a certain amount of aggression to "reach" the high note. It showed interesting dynamics that render the emotion and intensity that I was looking for. The 2nd version was a male vocalist with a grunge/rock style - Even though my audio track was basically whispering or singing very quietly, he was singing at his default/baseline level. There was a bit of aggression and raspy-ness in his singing, and when I reached higher notes, he was actually belting. As for AI being without emotion, and sounding lifeless? I do believe that the user of the DAW needs to breathe life into human vocalists, just as much as AI vocals. Human vocalists that I have worked with can also sound disappointing, even though they sound more dynamic in person. Adjusting volume automation and bringing out more of their breathing is really helpful.
him saying that this will never replace real singers by producers when creating an entire song for release is delusional. AI is going to totally disrupt the entire face of the music industry.. ..(there is no doubt about it).
They said the same thing about midi. They said the same thing about digital recording... A great, unique voice will never be denied.... AI can't change that.
Great video! It was mind blowing hearing what AI can do but it pissed me off at the same time thinking how people will create a youtube channel doing cover songs etc completely with an AI voice that sounds amazing...AI will perhaps eventually take over the human race I think it's too late to stop. For now I'll just keep on singing ❤ great voice you have btw. Cheers
Could be the contract is not indefinitely but say for a 10 year period the company is licensing the vocalists voices and after 10 yeara they need to renew the contracts.
Question.. if you generate the same section of voice twice, is it identical or does it have a "random seed" in it so you can double track the vocal to make it more stereo or thicker and it isn't just an identical copy of the first vocal that is generated?
When you add two voices, or even your own and an AI voice, doesn't it sounds slightly unnatural because of absolutely perfect matching pitch etc between the two tracks? I'd love if it had some options for introducing randomness
What happens if you pay for a subscription but decide 6 months later you want to unsubscribe? there doesn't seem to be anywhere obvious on their website to do this.
Looking for a (quality) real time voice changer for busking eg. I bought a Macbook air for this, and it didn't work, I know there is any others softwares, if someone knows a good one ...
You do know that the limits on what men and women can do vocally are largely cultural. If people think your a girl, people assume that you may sing alto or soprano, if they think your a boy exactly the same voice is discribed as singing in falsetto etc. I know a ton if people who once sung bass but now sing soprano, they can still sing bass but it takes a little practice to get back there. How to achieve all this is out there but its more our psychological inhibitions than physiological reality that inhibit us.
@@soundlearn that makes sense, because the rhythm is simple but the melody has some relatively difficult "passage notes" that makes it a little hard to execute perfectly. but i should say that you have very good phrasing/lyricism(means how you compose your melodic lines and put lyrics to them). i transcribed your song and played around with it and it surprised me by its neatness and potential. Normaly "producer youtubers" have enough skill to impress but not more. but i think you should lean more on making videos of creating songs. I am a composer by training so i am not on the "Producers youtube" side a lot. I went on a binge yesterday skimming all the producer channels and you and a guy called sevrida caught my ear because you have good instinctive phrasing and melodic variation in phrasing. After seeing your video about "doing youtube for a year and what you learned..." i thought i would give you this feedback, make more videos showing your skills of songwriting. Of course the clickbait brought me here and is equally important, but what made me stay is the song and the Spotify video essay you made. I wish you all the luck. hope you keep doing it for these next 2 years because you have good taste and good talent and i am definite that you will make it. just takes a little bit of time
According to the website, as long as you are a paid member, the vocals you create are royalty free and you can use them commercially, but be sure to double check depending on what tier you subscribe to.
@@bommaa6496 This software changes your voice to a profesional singer's voice, I'm looking for one that makes my own voice sound profesional even if I can't sing.
actually the original vocals were more interesting and better, the audimee one was very flat, and in the higher range pretty bad. For backing vocal maybe it's okay, but for main voice I don't think so.
These voices are royalty free. In comparison to some of the open source stuff that is able to model any voice, this company has worked with the singers to properly license their voices and these voices are allowed for commercial use assuming you are a paying customer.
is this supposed to be a good thing? having ai now be in the music we listen to? i get playing around with it, but mannn can't imagine the ways this could actually end up ruining a lot
Or !!! You could just get a female singer on the record ….wow there are a lot of them dying for some studio work …why A.i. what is the big deal ? How is that better then reality ??