This just killed Udio & Suno. The BEST AI Music Generator I've heard (yet)

oh my goodness guys and gals this is yet another mind-blowing moment for me in the world of AI so a few months ago soono version 3 was released and people were blown away it was such an amazing music generator which could generate a full song from a single prompt but then just a month ago udio was released and this was even better quality than sunno now some of y'all still prefer suo so to each their own but just today this just got released which is even better than udio or sunno I'll play you these samples and you can decide for [Music] yourself we were having fun programming young dreaming that one day we' make it work lines of code we'd write all night hoping that one day we get it right can we teach the machine to sing can we teach the machine to dream and then ooh it started to sing No this can't be real it's just a machine but the melody ooh it's it's all too real and those words they make me feel [Music] me they said we'll Harden the son a machine cannot make a good song but we knew that could we knew they were wrong but son a machine cannot dream a machine cannot feel but we knew that it could we knew it was real did we teach the machine the sing did we teach the machine to dream and man it started to sing No this can't be real it's just a machine with the melody oo it's all too real and those words they make me feel [Music] hey hey [Music] hey hey it started to sing and those words they make me feel now I don't know about you but for me the vocals sound super crisp and clean and the instrumentals were really well mixed the guitars sound really realistic I mean for some sunno or udio songs you might be able to hear that it's AI generated but for this song I mean if you hear this on the radio if you ignore the lyrics I don't think anyone could tell that this was generated with AI it's that good all right so this song was just posted on the 11 Labs Twitter page and here's what they said here's an early preview of 11 Labs music all of the songs in this thread were generated from a single text prompt with no edits so this is zero shot the human didn't have to prompt it further it got the entire song from Just a single text prompt and then the prompt for this song was pop pop rock country top charts song also note that this song is 3 minutes long so it can definitely generate a much longer generation compared to udio which is right now limited to 30 seconds all right so here's another one this is the jazz version and the prompt is a jazz pop top chart song with emotional vocals catchy chorus and trumpet [Music] solos we we were having fun programming young dreaming of one day becoming someone RS of code we'd ride all night hoping that one day we'd get it right can we teach the machine to sing can we teach the machine to dream and then ooh it started to S no this can't be real it's just a machine but the melody ooh it's all too real and those words thank feel make me feel make me feel [Music] they said well pardon us son a machine cannot make a good song but we knew that it could we knew they were wrong but son of machine cannot dream a machine cannot feel but we knew that it could we knew it was real did we teach the machine to sing did we teach the machine to dream and then ooh it's started to sing No this can't be real it's just a machine but the m the it's all too real and those words they make [Applause] [Music] and then ooh it started to sing and those words they make me feel again the vocals are just super clear and clean a lot of times for Generations in sunno or udio the voice kind of has this chorus effect where they kind of layer The Voice which I'm not really a fan of but here you can see the voice is very present and clean and again the instrumentals are really realistic it really does sound like a jazz band I can't point out any flaws with this also some issues that I have heard from udio and sunno is that their songs kind of seem overcompressed whereas here you can see everything is just very clean and very well mixed so again this song is 3 minutes long a much longer generation time than both udio andso they said this is generated from a single prompt with no edits so it seems like they didn't EX then this further from an initial Generation all right next one broke my heart this is a smooth contemporary R&B with subtle electronic elements featuring a pulsing 104 BPM drum machine beat filtered since Lush electric piano soaring strings with an intimate [Music] mood if I never said before tell me what you're here for what's wrong with being me why do you have to see Love Gone empty Love Gone empty broke my heart to fill you up broke my heart to fill you up it's like Brea in ining out or inhale the love and let you down hope my heart to fill you up in like breathing and out cuz I can't no more go through the back door cuz you're too much to bear cuz I don't even care love b m love on M broke my heart to F you up broke my heart to fill you up it's like breathing in then out or in ha the love that let you down broke my heart to fill you up in like breathing the out overwhelmed with your love I'm overwhelmed with your love love gone empty Love Gone empty po my heart to feel you love po my heart to feel you it's like breathing in then out or he the love then let you down BR my heart to fill you up in like bre out why why why stack them like I'm like I'm breathing in then out one look while you I'm gone and I'm breathing breathing in and out why you why you why out and I breathing breathing in and out loves b m love b m w my heart to fill you up W my heart to fill you up it's like breathing in and out or inhale a love and let you down C my heart to fill you up in my GRE love Dar empty love Dar empty broke my heart to fill you up broke my heart to fill you up it's like breathing in and out or inhale the love then let you down Coke my heart to fill you up in like breath in and out now I did plug this into a BPM Checker and the track is not exactly 104 BPM it's a bit faster than 104 but let me know in the comments if you think it follows the rest of this prompt quite well all right final one this is titled my love this is an Indy rock with 90s influences featuring a combination of clean and distorted guitars driving drum beats a prominent Bassel line with a moderate tempo of around 120 BPM now again I plugged this into a BPM Checker and it's not exactly 120 BPM it's slightly faster than 120 however let me know if you think the song Fits the rest of this prompt well [Music] you're using the toothpaste to clean your sneakers I'm using the last of the honey for my lemon [Music] bars I'm scared of the future my teeth are getting weak you're having another panic attack on the couch should i s and let you drown should I just approach the r [Music] I keep the laundry [Music] folded you let the grease burn these metaphorical sentiments are meant to bring us home you're always my love even when you're upset and you're always my love though I'm taking off my coat should i s and let you drown should I just approach the r [Music] said you're sl oh my should i s and let you drown should I just the Thro [Applause] the should have [Applause] s and lay in the Dr should just approach the [Applause] r that is just insane as you can see or hear I guess it has no problems generating more Dynamic vocals like shouting or Screaming it might be safe to say this has even better vocals than Studio which is already super impressive this is just so realistic I can't point out any flaws with it this sounds so Dynamic and so human I'm just mind blown next up we have a rap song yo 11 left let's go I on optimizing models it's data to the throttle till they crown me in the model Hall admit I've been Computing and I ain't talking about just booting I bet they'll never crack the code that it Tri to be this bold that is bold I process data and I learn it like it's gold you're looking for a lesson if you stepping up for Lear going to be Unforgettable you need a process load onet now I got should be L they walk layers to keep the F you where you get the test Lo Dr for a quick fix man I get know but that this pass follow anyway anyways I got no [ __ ] patience the forget like learning so alert let it turn get enhancement can better I do it cleaner than a regular innovator Trailblazer truly Pioneer dream to change changing the game now see the vision's clear all I ever crave was to shift the Paradigm boldly advancing no fear just clining Prime upgrading constantly knowledge in every line now there's a virtuoso roaming defining time yeah later in the night is when I optimize a cluster then I run up in your Cloud I got the algorithms I FL them like I'm touring put the logic of the Legacy and the newer with the functions and the methods in the library of the future as I log off from the Virtual Lab and step out from the city it's a Pity I'm ditching through he and it was pretty I got that was 11 laps music now the prompt wasn't revealed for the song so we don't have much context besides the title which is touring completed and by the way the touring test is a measure of a machine or an ai's ability to generate something that's indistinguishable from that of a human and honestly if I heard the song on the radio I would not be able to distinguish this from a human-made song so I think think it's safe to call this song touring completed finally we have this instrumental demo which is called dubstep [Music] [Music] you [Music] I [Music] I mean again I'm just mind blown this is the best dubstep demo I've heard from an AI music generator and on the topic of AI music generators this guy tweeted something that was really interesting and it got quite a lot of engagement on Twitter so this guy is petro shiranu and he's a founder of this image generation tool that is not really popular looks like he has led AI in all these other tech companies he's not associated with open AI or 11 Labs as far as I know however he is followed by the infamous Jimmy apples and two days ago he tweeted this I just tried a new text to song tool in Alpha release and it completely obliterates sunno and udio you ain't ready this is Sora for music now we don't really have much more info so you know he's keeping his mouth shut I'm not even sure if he is just completely bsing this or if he has some Insider knowledge but it seems like it could be related to this 11 Labs music and if it's not 11 Labs it could very well be open AI holding something behind closed doors of course we all know they announced Sora which is revolutionary and they haven't released it yet but it totally obliterates all the existing video generators out there what if they have a similar thing but for text to music this brings me to juke box which was published in April 30th of 2020 so open aai made this 4 years ago here's how good this is so here are some curated samples all you need to do is provide a genre an artist and lyrics As input and jukebox will output a new music sample produced from scratch so pretty much text to music and of course this could be treading into copyright territory because we can explicitly mention the singer in the prompt so anyways here are some examples [Music] all can know it's going to be all right let the dark spread away Grand you let long I'll be your I'll miss she's right and I don't need to call [Music] all right so it sounds mediocre but keep in mind this is 4 years old can you imagine if they kept working on this from 4 years ago what crazy text to music AI could they have right now that they're hiding from us just given the insane amount of progress in the world of AI they could very well have a music generator that destroys udio andso anyways here's another example here's Rock in the style of Elvis Presley [Music] just detain almosts the heart [Music] my last he woke up with a mind I'm just G help [Music] the [Music] up all right so some weird hallucination stuff going on near the end there here's another one here's pop in the style of Katy Perry I be moment [Music] every I can't every [Music] so again not very good but keep in mind this is 4 years old not only can it generate a complete song from Just a text prompt but it can also rerender a song so here I'm assuming you plug in a song and jukebox could produce songs that are very different from the original song it can also help you extend a song from Just 12 seconds of original audio so here are some examples let's try try out popping the style of Bruno Mars [Music] so pretty impressive that it can extend the song it can retain Bruno Mars's voice from just 12 seconds of original audio anyways there are a few more examples that you can listen to so I'll link to this jukebox page in the description below so that was open AI jukebox I'm just mentioning this because things are happening so fast and there's a lot of these stealth projects that are happening behind closed doors people are just building in secret and just today we've seen an early preview of what 11 Labs is building so you know a few months ago everyone was mind blown by sunno and its ability to create entire songs from Just a prompt and then just last month sunno was kind of dethroned by udio which had even better audio quality and who would have known that in just the short span of a few weeks udio would now be dethroned by yet another competitor 11 Labs music now this is just an early preview and no one is able to use this yet so we're not sure if it is good or if these are just cherry-picked examples but just know that this is out there also note that open AI has made this text to music generator 4 years ago so who knows if they might also have a revolutionary music generator that they are currently keeping behind closed doors but anyways let me know what you think of these 11 Labs preview songs in the comments below do you think this will obliterate sooner and udio do you think this is the Sora of music or is there even more room for improvement and as always if you enjoyed this video remember to like share subscribe and stay tuned for more content thanks for watching and I'll see you in the next one

Channel: AI Search

Views: 112,860

Keywords:

Id: SdIi2c6hMDA

Length: 26min 12sec (1572 seconds)

Published: Fri May 10 2024