Unlock AI Vocal Creation with Audimee: Elevate Your Music Production

[@LanewoodStudios]

So what do you think? Do you have any experiences with this already? You can register for a FREE test account at: audimee.com/?ref=lanewoodstudios

[@dccarsonmusic]

I think this is awesome! It definitely is a great tool to generate some background vocals. I don’t always have access to female singers so this could fill in the gap. That and it allows you to mix up background vocals with different voices. I could see this working to create a small choir. Great job!

[@LanewoodStudios]

Thanks, the results are very good indeed!

[@officialWWM]

You’re wrong. It’s crap! And lazy!

[@LanewoodStudios]

Thanks for your opinion.

[@Peter-gu9ph]

Wow this is amazing! All versions sound good but I like the RnB vocal the most!!!

[@LanewoodStudios]

Yes it’s very good at creating your own AI voices as well.

[@markdeboerfotografie]

Fantastische ontwikkeling voor het produceren van tracks. Dank voor het creatief aan de tand voelen van Audimee 🎤🤖

[@LanewoodStudios]

Ja klinkt lekker hè 😎😁

[@shabutelo]

Awesome! It's crazy where its going with all that AI madness haha

[@LanewoodStudios]

It really is!

[@stageleftguitar5887]

Wow…. Most impressive! I used this to see what it would do for a couple of layers during a chorus in a hard rock/alt metal song. I used the singers higher harmony tracks and the AI generated a nice falsetto I used Sebastian & Austin and even though they are more for pop/hiphop using those as layers was absolutely outstanding. They added a bit of extra width and depth to the mix. Really nice extra little “something”. There were a few audible artifacts by the time I got through heavily compressing & eg’ing , but it was not noticeable at all since they were not prominent in the mix and we’re just there for fill and texture. Overall, I was extremely impressed with what the free trial provided for my test. Thank you so much for this video! Having the ability to quickly add an extra vocal to a mix is a game changer! it’s a little scary what this technology can do but very impressed with the outcome.

[@LanewoodStudios]

Thank you for providing your experiences and great that they were so positive!

[@user-xr1ej8nx3s]

Bedankt voor de geweldige video's.👍

[@LanewoodStudios]

Graag gedaan!

[@SableWinters]

Excellent demo!

[@LanewoodStudios]

Thank you!

[@arupsircar6620]

This is really interesting working on AI.. but ur explanation is fab..n that's d reason my thumb is posted n the bell keeps ringing.. Spent 50 years in my musical career.. can't survive without Music.

[@LanewoodStudios]

Great, thanks for the support!

[@stefanhagel8496]

Great video! I personally wouldn't use AI for lead vocals, but let's say to double up self sung background vocals. Never the less, it's absolutely phenomenal how AI can affect music meanwhile. And I think we're only at the beginning.....

[@LanewoodStudios]

Yes this is only the start of the AI journey I’m sure.

[@MartinMartin-yi9to]

As a creative tool this is amazing. There is a chinese company offering you just type down the lyrics, add a melody line and they provide an AI vocal track.

[@LanewoodStudios]

Yes you're only limited by your imagination with these new tools.

[@ErixSamson]

Can you share the name of that Chinese solution?

[@Utilitymatrix]

Please share the name

[@MartinMartin-yi9to]

@@ErixSamson Dreamtonics

[@Emily_M81]

@@MartinMartin-yi9to thanks! Sheena sounds really good o.O and way cheaper than any Vocaloid lol...

[@Spineloro]

VERY NICE ! LOVELY SONG !!!

[@LanewoodStudios]

Thanks!

[@kayodeoluwafemi8955]

I wonder how long it will tools like Nectar 4's Backer to catchup with Services like Audimee as a vocal replacer. Subscriptions can work out expensive in the long run.

[@LanewoodStudios]

Yes they may very well catch up and then be cheaper.

[@CharlesK441]

Fantastic and Fascinating 😮

[@LanewoodStudios]

Yes it is!

[@Sotangy247]

Brilliant work.. Track should be released

[@LanewoodStudios]

Thank you. The track is by Austin Hull and has been ‘released’ as a Cubase demo project.

[@SableWinters]

Have I seen @LanewoodStudios in the @Cubase chat?

[@LanewoodStudios]

Not very often but yes I’ve been there.

[@the35thScoutGroup]

Intriguing. I wonder can I carefully record myself (voice not so good anymore) and get a good model? I also wonder about voices being captured by the source owners and perhaps used - do you inadvertently give away your rights? Fascinating stuff though!

[@LanewoodStudios]

Yes recording yourself (or someone else) and then making a voice model is the default use-case for that I believe. I just used samples because it was a nice test which worked out very well. The copyright and legalities of it all is some uncharted territory at the moment I think. Obviously you can keep the model of your own voice to yourself, but if someone has clean recording from your voice they could make a voice model as well. That's what's happening to some famous artists now.

[@SableWinters]

Subscribed

[@LanewoodStudios]

👍

[@opodendorf]

It could also be interesting to feed the model with a synthesizer sound, for example. Maybe this creates a vocoder sound?

[@LanewoodStudios]

Great idea. Haven’t tried that one yet. Maybe in the next video 😉.

[@JayRP-pk5de]

Awsome!... I need this... 😃

[@LanewoodStudios]

Great, use my affiliate link 😉.

[@Gamme03]

thanks

[@LanewoodStudios]

@@Gamme03 👍

[@alexissongsforyou]

Is there a place to go to get 10-15 min of dry vocal samples in various styles? I'm looking for male vocals in sort of a rough kind of raspy style.

[@LanewoodStudios]

Not sure. Maybe if you look for dedicated sample packs like I did for the Steinberg samples. Even just on Splice or LoopCloud. None of the available Audimee voices work for you?

[@alexissongsforyou]

Hi! Thanks again for another excellent tutorial! I wanted to see about emulating what you did with the female voice in the Cubase library, but for male voices, just for fun. I couldn't find any male library voices in my Cubase 12 Pro library. I'm also conflicted about the ethics of using AI singing voices ... if the singers whose voices were scraped didn't consent for the AI engine to use their voices to create AI voices, is it OK for me to use those AI voices?

[@alexissongsforyou]

PS: I went to Splice, Looperman, and one or two others ... it's surprisingly hard to find unprocessed male voices in general, and harder still with a specific style/ sound in mind.

[@ilyaorlov]

Thanks, great review!

[@LanewoodStudios]

Thanks for watching!

[@kintubeats]

Insane!! Great tutorial

[@LanewoodStudios]

🙏

[@ErixSamson]

The audimee use case is to rep^lace an existing vocal track with a new voice, but same performance, right? It cannot generate the vocal track from lyrics and melody / chords for instance, or can it?

[@LanewoodStudios]

Yes, it needs an original vocal. In my previous video I tested it with my own vocal even though I cannot sing :). See the link in the description or at the end of the video.

[@heartshinemusic]

Hoi, bedankt voor nog een tutorial. Ik switch naar Engels... What could've made this demo better is a more dynamic source vocal track. I mean with more different volumes and emotions. I'm curious how well Audimee holds up when you really push it to its limits: with soft (whisper) singing to strong high belting. In other words, testing the vocal range. There are some big pro's and cons for producers: Audimee exactly copies your input. This gives an enormous level of control! On the other side, you need to provide the algorithm with a vocal performance that is already really well sung. So for people who can't sing well, it doesn't offer so much. But for producers who can sing pretty good but don't have a great voice/timbre or a small range, this is groundbreaking.

[@LanewoodStudios]

Hi, yes you do need a pretty good vocal performance to get a nice AI Vocal. I did already try Audimee with another (slightly more dynamic) vocal of the singer of our band in a previous video, and also tried what it could do with my own vocal (whereas I cannot sing). The video is linked in the description and at the end of this one. So yes, there are limitations, but the current results are already quite staggering to me. Fijn weekend nog!

[@dandtintennessee7675]

I usually record Instrumental music but i do have a few song with lyrics. Problem is i cant sing worth a flip although i try.. this could be a good tool so my songs with vocals can come to life.

[@LanewoodStudios]

I tried that in my first video with Audimee. Check it out!

[@dandtintennessee7675]

@LanewoodStudios I'll look that up. Man I sure wish I could sing.. lol

[@joksimovic5573]

Hello. I have a question. Does the language have to be English or is it possible to have the AI ​​vocals also sing in Arabic or Indian? Thanks

[@LanewoodStudios]

I’ve heard that some people have had success with foreign languages so just give it a try.

[@joksimovic5573]

@@LanewoodStudios Ok! Thanks

[@PatrickSchouten]

Thanks. What will happen when you upload the same track to the same AI artist... will the created track be exactly the same or does it differ and this way you can create double tracking?

[@LanewoodStudios]

Nice idea, have not tried it though! If the vocal generation is deterministic it should generate exactly the same Ai vocal again of course.

[@Bradleybrookwood]

Why is it seem like every time I want to add more sparkle to my voice it turns my voice into a non-English voice when I'm singing in perfect English? Do the models sing in English or are they non-English speakers? Edit, also it would seem that in order to do really spacious hard super wide panned vocals you would have to use different voices you can't use the same male voice and it's so stupid.

[@LanewoodStudios]

No sure why you get that. My experience is that it retains the character of the original voice very well. So if someone has a bit of an accent you hear that in the AI voice as well. W.r.t. using it for background/panned vocals, I tried that recently as well and made a separate video about it: kzbin.info/www/bejne/m4TaiIKNjalme9U

[@Bradleybrookwood]

@@LanewoodStudios it's still a bit confusing to me. When you record backing vocals do you need to pick different voices so that the panning feature actually works or can you use the same voice with multiple backing vocals? Sometimes when I try to do the backing vocals with certain voices, sometimes when I use the same voice on multiple tracks sometimes it works and other times not

[@ВладимирСмоленский-н3р]

❤

[@LanewoodStudios]

👍

[@Emily_M81]

thanks for sharing! I'm still kinda salty about MP3 on any paid tier, though =/

[@LanewoodStudios]

😎

[@davidsiliquini5433]

👍👋🏻

[@LanewoodStudios]

🙏

[@hertzouttahell]

amazing what AI can do these days. Thousands of Weekend tracks incoming 😅

[@LanewoodStudios]

Ha ha ... depends on the rest of the music as well of course but yes, this track sounded quite inspired by that. I have another video in which I first explored Audimee and that music is entirely different. Much more stripped down.

[@pakipoli4895]

WONDERFUL!!!

[@LanewoodStudios]

Yes it is!!

[@leoekks3747]

Why you not test your own voice ower some " pro" vocal samples?!? It would be much more interesting.

[@LanewoodStudios]

I really cannot sing and I already tried Audimee with my own voice. Check it out in the video in the description.

[@MikeBrayton]

Why do I feel unimpressed? You're giving it the main vocal line, from the singer in Cubase, sung well, and it is generating a tonal difference, but the vocal is sung exactly the same. You still have to start with a really well sung vocal, in tune. So if you're a home studio person who can't sing, you got to find somebody who can. However, if you can, then you can create background vocals that sound different. And that is super cool. But one good thing about background vocals is that they aren't sung exactly like the lead vocal, giving you variation that sounds human and real and cool. It would be great if the background vocal generated here could also mimic those performance differences.

[@LanewoodStudios]

Hi Mike, thanks for your comments. Yes it gives best results when fed with a decent vocal to begin with. In my previous video I did check what it could to with my own voice however, check it out at: kzbin.info/www/bejne/jJjKq5Vsi8tgiZY And I cannot sing at all really :). The AI vocal is definitely more bearable than my own voice though, even though it even picks up on my Dutch accent. For backing vocals, yes you do get backings sung in the exact same way but with a different voice of course. I can imagine that you can sing different versions of backings yourself and then generate AI backings from those different versions. Or even if you recorded just a lead part from someone, you may still have other takes of the main vocal that can be used as source for AI backings. Another use could be that if you don't like the character of a voice that you recorded you can replace it by an AI vocal that sounds pretty good. On the example project I like some of the AI vocals better than the original. Anyway, time will tell what this technology will be used for.

[@MikeBrayton]

@@LanewoodStudios those are some very good points. You can use different takes of the main vocal to generate backing vocals! Or when I sing my own backing vocals, I can replace them with an AI voice. And wow, if it makes your own voice sound better... Now I need to go hear this Dutch accent. :) Thanks for the great videos.

[@LanewoodStudios]

@@MikeBrayton check it out!

[@newshoesbigblister]

But...the female artist who recorded the clips for Steinberg/Cubase - she may have provided those clips on the basis that they would not be used in this way without an additional fee or at least a contract stating where/how a likeness of her voice could be used. She's a fabulous singer, but we all know people will try to get away with using the Cubase samples to create songs to sell. IMHO, the ethics/rules should be much clearer on use of AI.

[@LanewoodStudios]

Yes it’s still a legal area that needs to be sorted out I think. Using these samples in songs and sell those songs is legal for sure. It’s just like with any other sample library. You can even chop them up and process them as you like. However you can now also use them to train an AI. In future, that kind of use could be an exception for use of the samples. Maybe some sample libraries already specify this.

[@Emily_M81]

I'm pretty sure they trained their Aston on Haley Williams (the demo audio is a Paramore song lol). They sound REALLY close, and if you don't listen closely one could mistake Aston for Haley; there are subtle differences, though, and the more I listen to Aston and Haley back to back it becomes easier to sort them out. But damn is it eerily close.

[@LanewoodStudios]

@@Emily_M81 sounds great whoever it’s modeled on 😉.

[@Lorenzo1950]

You just put 20 musicians out of work.

[@LanewoodStudios]

This technology is here to stay so I figure it’s good to know how to use it. Drum machines didn’t put all drummers out of work but it did have an impact of course.

[@dandtintennessee7675]

Or put 20 potential musicians to work.

[@MichelBarbaro]

It seems to me the only way to sign in is "Continue With Google", what is scaring. Am I missing something?

[@LanewoodStudios]

Yes I've also registered with Google Authentication. Not really scary to me because I use that a lot actually, but you could ask them if there's another way perhaps.