Mysterious AI Music Generator 2x Better Than Suno?!
Vložit
- čas přidán 31. 05. 2024
- What is this new mysterious AI Music Generator that some are calling a Suno Killer, or at least twice as good as Suno v.3?
Rumors have been circulating for a few days now, so I've decided to put my detective's hat on once again, for an AI Mystery Investigation. In this video, you'll not only learn what this new platform is, but we'll hear some samples from it, which I'll do some analysis on to see if it really IS a threat to Suno's AI Music Throne.
Plus, we'll hear some thoughts on AI Music from Jordan Rudess, the virtuoso keyboardist from the band Dream Theater!
LINKS:
Udio: www.udio.com/
Delirium from Coffee & Cigarettes: • GZA, RZA and Bill Murr...
Pianote w/ Jordan Rudess: • Jordan Rudess Hears Al...
Chapters
00:00 - Intro
0:27 - The Mystery Begins
1:38 - First Sample
2:38 - Sample Reaction
3:08 - Sample Analysis
3:58 - Who is this new Music AI?
5:43 - Second Sample
7:30 - Second Sample Reaction and Analysis
8:05 - Thoughts on Gen AI and Musicians
8:47 - Interview with Jordan Rudess - Věda a technologie
UPDATE: Udio has been released, and it's FREE: czcams.com/video/KxEbfVmI67o/video.html
Google: Udio is risky website, not a secure 😢
Couldn't get it to work. Song generation just keeps spinning with no end. Bummer. Suno is still working great though and spits out songs in mere seconds.
haters be like AI is overhyped. They have no imagination, and consequently will be scrambling while the rest of us were like, told ya.
100%. I hear stuff like that second example and I want to load it into a sampler and start going to town on it.
For straight up playing, I can't wait to generate isolated drums and play bass/keys/guitar over it.
There's SO much to explore and do here-- or even just "Bob Marley in an Death Metal Band with an EDM breakdown" why? I don't know! It sounds insane and I want to hear it!
You guys are saying that but 99% of you aren’t making money with these AI tools. You’re saying “adapt” but what services are you offering with these AI tools to make money? You guys are just AI hobbyists. You’re not adapting to sh*t.
@@TheoreticallyMedia only 100%?? That's like just 1X 😢
Agree completely. Learn or die. Innovate or die. Adapt or die.
Stopping this growth will be as pointless as the RIAA shutting down Napster in an attempt to stop music sharing and insisting everyone to buy CDs. Only this genie is way, way bigger, and has long ago escaped the bottle.
Who the hell cares?
Thank you, Tim. For keeping me up to date on developments in AI. Love your channel.
Wow! That second song sounds like an actual recording from the 60s!!
Seriously, right? That's where I think the current "lo-fi" sound of AI Music really works-- it makes it sound like it's coming off a dusty old record! It's so cool!
@@TheoreticallyMedia I was blown away by it. With approval given by the estates, I can see the revival of some classic acts with this.
@@eddysgaming9868 Indeed, and I think that artists should be given special access to creating music in their established style. So while everyone else can only approximate something in the genre, the band can have their voices and their song and even a particular song or set of songs as the inspiration for more. They can then polish it up and take bits from other attempts and make something unique but with far less work.
@@TheoreticallyMedia I was just going to type that. The lo-fi AI sounds work really well, even in Suno. The imperfections help them become so much better. Vocals were pretty impressive. Lyrics - it'll get there.
I crated a black sabbath song with udio and convinced pppl it was a "lost track" with ease so its good for "deep fakes" but not for modern creativity like suno v3. Suno is MUCH better.
Thank you for using something to remove background sounds in that interview at the end. Definitely easier to hear him.
Oh my! Thanks Tim for the usual standard of excellent info! As a prof musician, I have no idea what to make of this!
Oh, I'm with you. I'll say, I spent some time talking with Udio yesterday, and that back end part where they are working with musicians-- yeah, that's legit. They're really aiming for this to have two components. One is the "fun" version for most people-- but, also a "pro" version at some point. That will be more focused on assisting musicians in songwriting and composition.
Not too worried about it-- we have a lot of tools like that already. Everything from Drummer in Logic, to "Smart Chord" generators.
At the end of the day: No AI Tool can play a live gig as well. If it played the gigs I did, it'd fry from all the beer that was spilled on it! haha
@@TheoreticallyMedia I think you need a beer hat, you remember those?! But seriously, just like with the visual arts any of these tools along with a human in the loop can be wonderfully inspiring. However, it does beg the question if it's so easy to generate content like this will younger people be encouraged to become excellent at playing writing recording and engineering? And of course it only makes me think the pendulum towards organic live and in person performance will be the future of music and creating communities. Just like it used to be many millennia ago around the proverbial campfire! And that's OK with me!
Jordan! I'm not worthy! *Bows* Cool mini interview. Signed up for Udio. Thanks for doing the detective work. It was buggin the hell out of me not knowin.
This is backed by Common and other professional musicians and I have to wonder why.
I am a professional author and also a composer. Up until recently I typically lay down my beat track and bass line, add my fills, the rhythm, the hook and then play through the melody on whatever instrument I'm using. Then I would hire musicians to fill in the rest of the track or use my sound packs that work with whatever studio software I was using at the time. With this, I do not need anyone any more. In fact, my fans do not need me. They can generate as much music they want to hear on their own (when this is live and licensed for distribution) as they want, when they want. No longer do they need to watch for my next EDM drop or jazz track - they can do it all on their own.
Same for my publishing. I used to write and sell four books per year. Now with AI someone reasonably versed in prompting will soon be able to spit out a complete novel in minutes, and based on characters they want, a storyline they want, locations, genre and everything else. We no longer need authors or musicians.
I am rushing to release my latest three book series and to finish my in progress three book series because in a year or so, I think most readers will be logging into Amazon and selecting the book they want generated from a series of drop down menu choices. Same with music.
Creativity is passe.
Nah
Suno v3 instrumentation only is actually very good. There are a few key genres it does extremely well. In fact it's to a point I officially have hired someone to generate my full credit amount for me each month so I have a consistent backlog and idea folder to pull from.
Synthwave and punk pop is fantastic in Suno. It totally nails those two genres SO HARD. Emotional folk rock is also tremendous.
The last time tech blew my mind this hard was the first time I tried VR. It's startling how good this AI stuff is already.
Classical music and Jazz are both excellent in Suno Instrumentals. I wish the vocals could be as good.
@JimmyNuisance I, too, have been impressed with the Punk Pop, haha. It really nails the melodies and integrates a top line into the backing track.
I used to produce professionally, and honestly, it's just a perfect idea machine. People are mad at it, yet it's created 3 more jobs at my studio, so I'm definitely loving the process so far 😝
One area it still struggles is Disney or broadway musicals like music but hopefully soon lol
Suno is great at interpretting Post-Punk.
Tribe Called Quest.....to my ears...
Totally what I’m hearing. And now I’ll be listening to The Low End Theory later tonight!
@@TheoreticallyMedia That's what's up!!
I got the same message about contacting referrer, look forward to them giving access. Will be keeping my ear to the ground, as well as checking out your new videos 👍appreciate your content do much, and learned so much from them too, thank you 🙏
Naw, I just gave Udio a whirl, and Suno remains King, imo.
Yep, Udio pumps out a lot of lame stuff and takes 10 minutes each.
Udio has more humanized voices. But does not know how to structure and it's like you have no control, even giving as much info as you can it gives you what he wants, not what you asked. Is like going on a terrible barber shop. 😂
@@LucasdeFarias Suno has way better voices too. Udio just runs their generation longer for crisper sound.
Cannot use udio. It says "Please contact your referrer for expedited processing" after I logged in with my Gmail
Great video. Best coverage Ive seen on this. Truly impressive. Definitely a step up from Suno v3 in many reaspects. The multilayered vocals on the soul track were amazing and the natural swagger of the singer really hit home. Ive got a good ear... and that was sounding just right. Dont want to think of it as a Suno Killer... but as healthy competition in a really exciting field. The more new systems come online... the faster the whole thing will evolve. Thats the real take away for me. Strong competition will just supercharge the development of AI music... and thats a great thing for all of us. Really exciting news... will be watching this space closely. Thank you for this amazing breakdown... appreciate your thourough approach.
I can't believe you got the wizard himself on an interview! Been a fan of dream theater since high school (so basically, Images and Words). You da man, Tim. 👏🏽
Oh and ya, Ai is probably scared of Jordan Rudess 😂 he's that insanely gifted. 🤯
I was just about to say "lets see how it does with prog rock" and you bring the Wizard into the video! Great stuff
currently I'm using Suno v3, but definitely gonna try this one..! Thanks Tim!!
You did meet Jordan Rudess !! He's an incredible musician and has a great spirit, dream theater is legendary ! I love that he sees the bright side of AI and i'm also not surprised ! :)
Do you know if Udio's output is 190kbps mp3 too, it's a quite problematic format unless you have an idea how to " Upscale " audio into a better quality 😎 Thanks for the video and amazing content as always
Jordan (I’m happy to say) is one of the nicest folks I’ve ever met. Musician or otherwise! We hung out for well over an hour talking about music, tech, and AI. Oh and of course, nerding on prog rock!
So, I’m presuming Udio is going to be Mp3. I doubt at this point any of them are doing wavs.
Although, I have been thinking about reaching out to a fellow YT’er I’m pals with who is really good with mixing and mastering to see if we can team up for a tutorial on how to “clean up” AI songs.
@@TheoreticallyMedia Man you are amazing ! Very happy for you that you had a chance to hang out with a modern day legend !
For Udio, I have lots of ideas on how to clean a song or even reworking it but I secretly wish some insane tool that would be trained on several songs being degraded in quality so it understands as a diffusion model how to make a low quality track sound better (but it's probably not that easy hahah)
It is getting exciting!
It really is! BTW: Great channel! I grew up on O'ahu! Not a day goes by that I don't miss Zippys!
@@TheoreticallyMedia Mahalo. I enjoy your channel as well. The next time you feel homesick, feel free to pop in to the Kilauea Live Stream and say aloha.
Love these vocals!!
It's pretty remarkable. The second one REALLY floors me!
@@TheoreticallyMedia i had tears in my eyes.. wow...
Can you provide the link to that image about suno artist names style for "tricking" suno to do a similar voice?
posted up on the Patreon-- you should be able to access for free though!
www.patreon.com/posts/suno-music-101961487?Link&
@@TheoreticallyMedia thanks
You still hear the kinds of low-resolution audio artifacts found in Suno around the 5:50 mark in this udio demo song.
Yeah-- those will get hammered out soon enough. I actually think they kind of work in favor for "vintage" style generations. Almost sounds like its coming off a dusty old record.
@@TheoreticallyMedia The problem (as I see it) is these systems are not trained on an individual track level. It's incredible they are as good as they are, but if systems understood how the separate tracks are mixed and mastered and how they then fit together, we'd be seeing music generation taken to a whole new level of control.
I think tough times are ahead for musicians working in the advertising industry. They will use it as a tool at first but will be replaced by Ai eventually.
I guess just like most jobs in the next few decades
Yeah, that's been happening for awhile. Stock Libraries and whatnot. It isn't the worst thing, most musicians weren't making very much off that to begin with. Although, I do understand that when you're living the musicians life, every stream counts toward something.
That said, Musicians have gotten screwed over since the first paying gig. haha, they're a resourceful breed who will find another income source. Source: Am a musician.
@FlipFinderzhow do you adapt to this though? The thing does the entire artist, musician, mixing, engineering. There’s no adapting. It’s game over.
Idiots like that don't think more than 5 seconds ahead @@lukewilliams7020
Right now it still sounds pretty mediocre. These two shows are probably the best they got out of 100s if not 1000s of attempts. Also Suno is still better IMO.
9:38
So he sees it as a tool of inspiration....... which is great!
Totally! He's pretty deep into AI. Hoping to do a follow up with him shortly. Apparently he's going to MIT soon to work with their engineers on some more AI Music research. Really, really nice guy as well!
Tool of inspiration or not, the work for musicians will shrink substantially very soon. When people can generate music for their youtube videos the same way that they generate thumbnails for them, musicians will see a pretty steep cut..
Why would anyone pay a musician who is inspired by AI when we can just generate exactly what we want with that AI? If I was ever doing youtube content I would 100% be using AI for all my thumbnails. And if I didn't already know how to play a bazillion instruments and produce and all that stuff, I would also have used AI for the music. Actually probably still would use AI for lots of the music in my hypothetical videos. It's made in seconds and sounds adequate RIGHT NOW.. Just imagine 2 years from now.
It's cool that people are inspired by this, they can be inspired to make music for free in the coming years. Music no one will pay for or ever hear.
Great for hobby musicians with no aspirations. Totally going to destroy the people who make money off of stuff used for background music or the people making music intros to content and stuff like that.
Graphics artists and musicians will have to just accept that more of them need to see art creation as a hobby, not a source of income. Very few people out there will be able to really improve upon the stuff that's generated in 2026 or so, and they also won't need to. And more devastatingly, normal humans will not be able to tell, nor will they care if the music on a streamers stream or a tubers youtube video is AI or from a website that pays musicians to write royalty free music for those libraries.
When AI started kicking off 2 years ago I was pretty positive. As time has gone by it has become pretty clear that this is going to completely wipe out a very specific sector of the music industry instantly, and then over time gobble up most of the rest.
It's like the home studio revolution of the 90s and into the 2000s that caused so many more artists to pop out of the blue with self produced albums.. Only this time it won't take any skill what so ever, so the market is going to be totally flooded with quite decent music. Adequate for professional use, and nearly free to generate. Why would you pay a graphic designer to do your thumbnails and a musician to do your intro when you can get substantially better results just from writing a few words into a box? You could make a song with fewer words in a prompt than you would have to write to a guy on Fiverr... It's so ungodly easy. The dude you write to on Fiverr will also most likely just take those words and feed them into an AI by 2026. No fixing needed, it'll just generate a bunch of songs and then he picks one to send back.. Charge 30 bucks.. Onto the next.
But hey, musicians are gonna be real inspired. They're also going to work full time at mcdonalds with their hopes and dreams dashed by AI.
So many people out there with decades of experience in music just going "damn... I guess that's that.." And they're right. What's done is done.
I was never one to believe the line that "people will prefer human made art", I don't think that's even going to be a factor in 2-3 years. People have been extremely unimpressed by musicians for a LONG time already. Back in the 80s you could get laid if you knew how to do a pretty decent solo. These days, nah. No one idolizes musicians they way they did. We're easily ignored and pushed to the sidelines..
But man, so inspired...
@JimmyNuisance
👋 🫂
I'm hearing you and what you write makes a lot of sense.
But I'm sure most of us here realize that this fate for those mentioned is inevitable and if one does realize this, they ought prepare for that inevitability......for their own sake.
*"Technological achievement" adores no Hurrian Hymn No. 6.*
@@JimmyNuisance Musician/Producer here.. I think you're spot on
Rudess nailed it. You mentioned before about (some) musicians supporting this AI growth, I've seen the opposite mostly. Such as the open letter warning against the “predatory use of AI” in the music industry by some 200+ musicians (Nicki Minaj, Katy Perry, Billie Eilish, Stevie Wonder, J Balvin and Jon Bon Jovi).
I could write a dissertation on this, even going into socio-economic issues regarding art in general, but I'll stop before I even start, other than to repeat what I first said. I agree 100% with Rudess.
Yeah, I don’t want to get too deep into the woods on it either, but I’ll say: that letter makes no sense.
“We hate AI. We love the technology that AI brings. We don’t want anyone to use AI”
Most of those artists are basically mini-empires. At the very least, they’re companies whose main source of income is the artist IP.
I get it, I really do. There are little mini economies built around these artists.
I’m not saying Jordan and DT are by any means unsuccessful, but also…they aren’t Taylor Swift playing global stadiums. They’re relatively working class (high end working class) musicians. So, for Jordan, it makes sense to lean into this technology to see what can be done with it.
I've written just over 1000 poems, but never had any way to convert it into music. I am incredibly excited at the prospect. Even if it gets lost amongst the sheer volume of generated music it should be a fulfilling venture for myself!
That makes me super happy to hear. And hey, I’ve made tons of music that only exists on my phone. It’s such a great feeling to drive around and blast something that you wrote. Can’t wait for you to play with this!
Suno gives you 5 songs a day for free. Enjoy your musicalised poetry!
exactly the same with me. I finally hear my own words. Awesome.
I'm in the same boat. 1000s of songs so 5 songs a day isn't enough also I end up doing all 5 on one song just to hear other options 😢 gonna take me 3 years at this rate lol@@imusiccollection
@@leestrz4153 udio is near unlimited at the moment with 600 a month. Happy creating!
As a musician, this dope. I hope I can generate individual instrument tracks somehow. All I know is soundcloud rappers are going to go NUTS
what do you mean by SoundCloud rappers? they will enjoy the tech or opposite?
you should be able to prompt things like "Jazz Drums, Drums only" or "Death Metal Bass Song, Bass Only" to basically get stems. I've gotten some good results with Suno, so it should hold true here.
Oh, and yeah-- those 3rd tier beatmakers on beatstars are done. The ones with talent (who also know how to mix and master) will be fine.
@@TheoreticallyMedia No. How would they be done when they can prompt high-quality stems now....? Every beat maker is technically done. Also mixing and mastering will fall next via a generative A.i. Brother, EVERYONE is done.
@@user-xedwsg You're right. People who think the human element is important, and that the top tier beat makers will survive this due to their skill in making awesome beats.. They're gonna be surprised to find out that by 2026, top tier beat makers will be outmatched by AI with ease. At that point the only thing those beat makers will have going for them is their name. If they have some street cred, sure, they'll maintain some relevance.. But that's going to fade pretty fast as more and more AI generated music rivals their skill. Why pay a top tier beat maker thousands of dollars when you can spend 5 bucks generating equally good beats without ever leaving your home? Like why? Why would anyone do that? lol
Everyone is done. Everyone. Today, AI music generation is as bad as it is ever going to be again... Like a year ago, AI music was just plinky plonk sounds in a scale, kinda played at random... Today it's making super catchy songs with great hooks and melodies and choruses.. I spent the last two days messing wuth Suno, and that thing is god damn incredible. The fact that there's already something far better out there is NUTS.
@@JimmyNuisance I agree with you 100%! In a few more iterations of this tech, EVERY BEAT MAKER ON EARTH WILL BE OUTMATCHED BY A.I.
And the wildest thing with A.I. is that it spits out a beat in 5 minutes. FIVE MINUTES!!! I could make a whole album in a day, a whole catalog in a week! I dont think people understand how wild things are going to become.
And yes, why pay a beat maker for any ANYTHING anymore 😂😂😂
The 15kHz cut is because it has been converted to mp3. Just studied in music school
Ah, that makes sense. I vaguely remember that from my more audio-centric days. But, I'm also old, and came up during the Analog era! Haha, my first home recorder was on a Tascam 4-track. Which, I find endlessly hilarious that the LoFi crew lusts after for that "old school" vintage hiss and warble. That was all stuff we'd try to get RID of back in the day!
Awesome work!
Thank you so much! Really appreciate it!
Good video. Thanks for checking it in the DAW for us. The vocals got me more than anything to tell you the truth. It _looked_ like they've done the old turn up low end and mid-highs for perceived fidelity. It's in stereo, and my guess is they'll turn up the LUFS for good ol' loudness wars. Immediately, that'll sound better. The vocals were something else though, especially in the 2nd sample. Th link is setting off the computer virus thing by the way. Might just be me. Great stuff. Thanks. 👍 Oh, and not Tupac, no!
Muchas gracias por la lectura!❤
It's fun to play around with
Definitely Tribe feels there. Thanks as always sir!
Wow, that second track is amazing
I really like this tool, I never did an AI songe before and was able to figure out the basics of this pretty quick.
Anyone in the music industry that is fearful of these tools here is another way of looking at it. If I was a musician I would insert my lyrics into Suno and render out different styles, then mix the version I like. Remove the vocals with Replay and then re-sing the song or create a voice model of my singing. Separate the stems and remix it. Congratulations you just created a new song in a few hours. If you're not a musician, make the music, create an Ai persona with any Ai image tool. If it blows up, then hire a real musician to play it live. Everyone wins. This is the future! It's time to take back control of the arts. We will not need giant abusive labels anymore.
That's a lot of pointless work. By 2026 you don't need to do any of this. The song you get will be good enough, and what needs to be fixed you can do within the AI tool itself. The editing functions in AI music generators right now are abysmal, and that will change. Once it does, a total rube will be pretty much as good at making music as a musician with decades of experience... And the listener won't care. Today, if you hear a jingle in an ad, some musician was paid for that. In 2026, that's going to be FAR less likely to be the case.
This is going to completely obliterate the VAST MAJORITY of the music industry. The only ones that will be relatively safe are physically attractive women who can maintain fame. Influencers will start releasing music that will be as good as whatever Nicki Minaj or Taylor Swift is making. And certainly better than the crap Beyonce is making. Fame and beauty will be the main driver in terms of big artists.
Jazz and blues club musicians might survive. That's about it.
The music industry was already in a bad place. This is the end of an era. For the past 2 days I've been messing with Suno, and I've "made" songs that are already better and more catchy than most of the stuff on the charts these days. Making your own lyrics in that tool lets you create unbelievably specific songs for very specific purposes. These tools are super primitive today compared to what they'll be 2-4 years from now.
I think we're at least one more generation away from something cutting edge production wise. I am rediscovering Leon Russell - somehow except for AM airplay in the day I completely missed him as the artist - cause I did not understand his songs. Give a listen to Magic Mirror to see what I mean. There's also this incredible production value layered over his super-talented voice and piano. Wow.
Good Grief does that sound really, really good!
I keep listening to that second example over and over...it's wild.
2:12 OMG!! I dont even care who it sounds like, it sounds LIKE A MASTER RECORDING!!!
Well... its not. Probably if you listen on a phone or similar, the summed output gives an impression of a retro record, but on critical listening the mix is all over the place in the stereo field, with vocals crudely slapped on top. I sure love the vibe - hope its just one example, and it will get better.
@@Uncabled_Music it's a recording of a recording
@ouroborostechnologies696 Suno also has those "60s wild stereo" mixes when it draws from the corresponding materials, but the model is better at fitting it all together. This Udio example has better timbre, but is not as accurate in this example. Suno is mostly tidy, just the timbres are still on the low-fi'ish side a bit...
Color me interested. Hope this hype doesn't come too early and we'll have to wait 6 months. :D
They've got to be pretty close to launch from the leaks. I did read someone say 4/20, but I'm not sure if that was a stoner making a joke! haha...Like, why not 4/19? Or 4/21? I'm just suspect of that date!
@@TheoreticallyMedia Well - it's not unthinkable - i heard of car manufacturer who made their cars S 3 X Y :D
I mean i'm generally down to give other AIs a try if they perform well. Suno is really exciting to use, but finding the right samples (maybe i'm just picky) takes a while.
Competition is good. Keeps people on their toes.
Second one is crazzyy!
I've made some straight bangers with Suno lol. Mostly my own lyrics, and mostly one song with many options for genre to help me decide what to record later IRL.
the Jazz sounded suuuuuper good!
I RAN to sign up for beta, I now get this message, "Please contact your referrer for expedited processing" It says my referrer needs to verify me and then I can return with my special code and enter. Any idea about all that bizz? And thank you for a great share.
Wow. That second song. This is crazy.
always love the references here and there! (especially the zoolander one on the cover image 😜)
Haha, I had a friend years back that would ALWAYS do the "Blue Steel" Zoolander look in every picture you took of him. Totally unintentionally too-- like, that was his "photo face"-- I was attempting to channel it here!
Holy freaking heck the 60s track!!!!!!!!!!!!! Game over man. Game freaking over for music producers and artists.
I think the musician you interviewed probably doesn’t realise how advanced AI actually is and will become. He doesn’t get it. That’s why he said what he said.
Oh, trust me-- he does. He's working with MIT's AI Music research, and has released a few AI music apps of his own. He totally gets it.
@@TheoreticallyMediaah ok interesting. Then that therefore makes his take on it very very interesting. In some respects I think because he is an established musician he probably doesn’t feel he needs to worry about AI as he already has an established network and fan base etc which will continue, but for those starting out it will be even harder to break through than it is now I suspect
I'm not a musician but honestly as a listener I think musicians have absolutely nothing to worry about I'm already hearing ai music, just be uniquely you and market your music, millions of people already make music and a lot of them are good but without marketing its just another good song on the internet, so ai is not going to do anything outside of that norm there will still be that marketing aspect of it rather you make your music or let ai make it
Exactly the music industry hasn't been about the actual music for decades now, it's all about promoting your image
Music made for the charts will be safe for a little while, because that's mostly just celebrities fighting over fame.
Music used in the background of anything will be AI, so all those artists are gone.
Music used in ads, AI.
Music used for youtube intros and in the background, AI.
These used to be things people had to pay another human being for. That's going to end. Why would I write a mail to a dude on Fiverr and get him to make me a song I'll get 3 days from now when I could just write that mail into a prompt and get the song in 15 seconds?
Thumbnails for youtube vids and pictures in magazine/website articles are already done by AI more and more every day. There used to be people who got paid for that work, now they aren't. There's going to be no difference with the music.
An incredibly small sliver of the music producers out there are making music for the charts. The vast majority of music composers and musicians are making music for content like ads/online content creators/all sorts of film/corporate stuff and so on and so forth. That's the MAJORITY of the working musicians. Their jobs will dry up. They'll have to do something else in life, their entire sector will disappear when someone can make an adequate piece of music for 2 dollars from the comfort of their own couch while they're watching TV.
With midjourney I can make art that would cost me IMMENSE amounts of money if I had to get an artist to do it by hand. The stuff you can make on Midjourney would demand a very skilled artist to replicate. Very skilled artists cost a lot of money... OR I can get the same result for 0.3 dollars through midjourney. The cost difference is likely to a factor of multiple thousands. Not paying a good artist a thousand bucks for a piece of art when I can pay 0.3 dollars.
AI will wipe out entire sectors, and I just can't fathom how this isn't obvious to everyone... All the visual arts people called this a year ago, and they were right. CZcamsrs use AI for thumbnails now. The value of a Fiverr visual designer is now gone. No money to be made there.
Zero skill needed to make great sounding music by 2026. No editing needed, it'll just sound good. If it doesn't, generate another.. and another.. and another.. and another.. You can generate a thousand songs before you even get close to the cost of hiring an actual composer. Same with art, literally thousands of images generated before you reach the cost of an actual human artist.
The writing is on the wall here... As cool as AI is, it's not JUST cool - it's also a tool that eliminates the need for trained artists. The people who benefit are the people whose creative skills were lacking to the extent that they needed to hire artists. For artists, this is pure loss.
@@JimmyNuisance How many people actually make a living making songs on Fiverr? Can't be more than a couple hundred people at the very most, for most it's a side gig and a way to make a little extra money, making money in music is already next to impossible and yes this will unfortunately wipe them out. As for visual artists yeah I feel really bad for concept artists and people that do work for magazines etc. But oil painters will continue to flourish, I continue to follow people on IG that paint oil paintings and have huge followings and the fact that digital art has already been mastered by AI has not put a dent in their sales or the likes on their posts, because people enjoy watching the artistry of people that can use physical paint. But yes, sadly for digital artists it is game over.
@@JimmyNuisance you are right. It is also worth mentioning all the companies making plugins, midi controllers, instrument libraries etc. The demand for all that will drop drastically as thousands of mediocre music producers/beatmakers will cease to exist. When talking about music ai people tend to think about the top charts artists. But there are millions of producers, beatmakers, session vocalists, mixing/mastering engineers, session instrumentalists etc who make living off music. And they are in danger here.
@@cottoo1 There aren't millions of people making money off music. There are only maybe a couple thousand people in each category you mention who actually make a living
My first concert was Dream Theater on their images and words tour. I remember tickets were either free or highly discounted with a toy.
...I wouldn't be surprised if more Suno competitors pop up soon...
(how many good image generators are there now? 100? )...exciting times!
I like Suno better
my question is do I need to make the music first and then add the melody? Another question about the song I wrote is how to know if it fits into a genre, how do I do it? I'll test it until I find something that works, I really like your kiss video!
Time for a big law firm or government to sue and force Udio and Suno revealing the training set they used for their machine learning.
Yeah, I mean I get it- but also: there are plenty of other countries who do not care about training data sources.
So; there is a question: do we (the US, that is) hamstring progress on the training data issue and lose global market share?
And while yes, that might make sense for a little AI music app, but that also extends out to everything else in the AI space.
@@TheoreticallyMedia there are off course plenty of countries not obiding to international and local laws collecting and mining data etc. We dont need and cannot accept that californian investors and big tech (read Amazon, Uber, Google, Facebook, apple….) behave like corrupt nations just because they have loads of money. Go in, regulate and equalize. In the long run it will help humanity.
Good stuff! I went to udio via your link. I entered some basic info then got this message on the screen: Waiting for Verification (next line) Please contact your referrer for expedited processing. Any clue what I should do on that? Thanks.
Sorry, it appears that they haven’t built that front door yet. Remember, we aren’t even supposed to know about this website yet! I’ll keep an eye out and let you know when I hear more!
@@TheoreticallyMedia All's good. I got in. Pretty amazing.
What a joy to have AI replace a lot of the boring parts of making music. For me making music has always been about finding those nuggets by listening for something cool. Same process with AI music creation but way faster. I'm having a blast right now!
It is still sooo awfull and fake. Only noobs fall for it. AI should work as a plugin during production. Nothing more since it is not more
@@RicoLee27 I respectfully disagree, sure you can hear flaws but this is wonderful, and it's only the beginning.
Udio v1 is impressive to me, especially because I wouldn't have been able to showcase my lyrics in any other way.
2:02
😲 I'm finally really impressed by AI audio.
Things are moving fast!
I mean, this is really pretty insane! By the end of this year? Yeah-- Spotify is in some real trouble!
@@TheoreticallyMedia
It takes a lot to impress me; I'm definitely impressed with the quality of Udio...... much more so than Suno.
But the tide of competition usually raises all AI boats......... even those made of Rice Crispies! 😉
So if Suno is driven to better by Udio, that's terrific!
@@TheoreticallyMediayep. People will be recording their own music or creating it. People will be adding it to their playlists &remixing it any way they want. All royalty free. Actually the music industry might be in trouble… 🤔
this one actually makes modern hip hop in the style of modern artist (not saying names) this is leaps ahead
As a hobby musician, I enjoyed competing with other musicians. But competing against an AI is no more fun for a musician who has been working on his song for weeks.
Can you do one on audio remixes? Especially when swapping vocies of singers and changing styles. Say, some pop song to a heavy metal one. Thank you
unsure! Hopefully we'll find out soon!
@@TheoreticallyMedia cheers!
Hi!! Can't log in! message: Please contact your referrer for expedited processing.. and nothing happens. U know why?
To my jaundiced ears, the backing on the first track sounded as if it was fighting against being held down by a ton of bricks. Second one worked better - sound alright on an old transistor radio sitting on a windy beach after drinking too much beer. Interesting things to come though, hopefully by helping musicians to head off in new directions to avoid being swamped by a million me too tracks.
I’ll say, I’ve been slipped a few other tracks by my sources (can’t share them or they might get in trouble) and so I’ve heard tracks that aren’t crushed by compression. And I’ll say, the output off platform is a LOT better.
And agreed: the quicker musicians can get off the Spotify chain gang and move into another direction, the better.
@@TheoreticallyMedia
And you think this won’t flood the market even more? Why would venues even pay for bands when an AI music subscription now will be in the future.
I don’t get how a musician can look at this and go, NICE, I don’t have to know any music theory or instruments now! And think that’s a GOOD thing. Please help me not feel god awful depressed by watching this.
Currently as a musician, I don’t see how this could be good at all.
bro , your not playing by the rules... title should be...
SUNO KILLER!!!!!
Haha, I thought about it-- I really did! But, man-- I just can't bring myself to do it!
@@TheoreticallyMedia
Rules are rules...
AI world its 'killer'
Debate world its 'Destroyed!'
😜
SHOCKS
@@TheoreticallyMediathank you for not lol
This is amazing. I tried the link to UDIO. Please contact your referrer for expedited processing??
Yeah, that's not surprising-- I mean, technically we aren't even supposed to know about it yet! It might not be set up as of yet!
I prompted a song like A7x little piece of heaven with brass strings and woodwinds, like Tim burton, or New orleans horror Jazz, It gave me a song that sounds like what a7x early days would do, even had M.Shadows sing it, then had an awesome breakdown where the vocals switched to Myles Kennedy
So this is definitely a subjective thing because based on those samples, I think Suno is better.
Honestly, I don't think you could accurately judge from either of those samples because old school hip-hop is not meant to sound like good quality audio. It was based on scratchy samples. As for the vocals, I've heard hip-hop vocals on Suno that were just as good or better. The second song didn't really sound all that great to me either in terms of audio quality. The instrumentation might be a bit better for that genre, but I wouldn't know since I haven't heard much from that genre on Suno, nor am I interested in nostalgic music. I've actually heard much better vocals with Suno v3 than those on the second song. These had a lot of artifacts. And to my ear, Suno vocals are usually well mixed. We'd really have to hear more genres in total, and more current and popular genres to judge.
I think a lot of people who only fool around with Suno a couple of times on the free version don't actually get a good sense of what you can do with it if you spend time with it. As for Suno is also independent stereo tracks and hasn't been mono only for quite some time. A lot of people do a couple of generations and just toss it aside because they didn't like those generations. I use Suno exclusively on my channel. It's the main focus and is gearing up as a stream "online radio station" with only AI generated music. So I'm using Suno daily and generate numerous times while building one song, making decisions on clips, finding the right prompt combinations per each genre I generate, etc.. Suno does have plenty of issues, but I'm willing to bet this one does as well.
AI generated music going to eventually reach the same quality as human creations in terms of both audio quality and song quality. I've been a full-time musician and composer for media for many years and initially was very "salty" on AI music. But then I started using it and realized that it's not what many people think. A musician can do so much with these tools, included those that are complete generators. I kept an open mind and decided that diving head first into Ai music generators is much better than crying in my coffee about "the end of creativity", which isn't true at all as many will eventually see. Do it now, or be left in the dust.
Agreed on all points! I thought Suno v3 was only doing mirrored channels, so not true stereo? Like, you weren’t getting instruments panned right and left?
But, to your larger point: I’m by no means calling this a “Suno killer” or anything like that. I had a bit (that got cut) talking about how Suno is already working on v4 as well.
And yeah, as a musician myself, I LOVE all this stuff. Just for ideation, or doing stuff like generating a track, then converting to MIDI- I mean, that alone is gold!
@@TheoreticallyMedia - Thanks for responding and I really like your videos and coverage on where AI is going. I'm glued to this stuff.
Suno is panned stereo as of November 2023's update announcement. You can see it in the wave files and hear the panning.
Definitely great for musicians. I think this stuff is going in a very positive direction with stems, possibly individual tracks and the ability to give chord progressions, keys, etc. At some point, virtual instruments will come into the mix as I don't imagine sample instrument and sample loop companies are going to roll over and die. There's probably something in the works with companies like Native Instruments, etc. I've been trying to find out, but no leads on that yet.
yeah the ability to prompt chord progressions would be great :)
Amazing. sounds like Tribe Called Quest. clicked the link and it says contact your referrer for expedited processing. So I will wait and hope I get an email soon.
1:06
Nice!
Haha, I was thinking about you when I was generating that! Louis = Ships made of Food!
(I gotta work that gag in more often!)
@@TheoreticallyMedia
You know I'm a big fan of your AI galleon ships made of food images / videos ...... they are amazing!
Filmmakers are totally going to use this instead of licensing existing tracks. You could even add words that match the video content while getting the exact style you wanted. You could possibly even get lucky and have one of the tracks get on the pop charts.
Meanwhile composers are using sora instead of paying filmmakers to shoot the movie.
But filmmakers will soon be replaced with AI too. And many more jobs.
Sound quality from Suno is way better...it is noticeable by a lot..
I'm just waiting for this sort of tech to go multi track and be able to go into my daw. Then we'll be talking.
Same! I’m wondering who will be the first “Splice” of AI. Or, maybe even someone figures out how to turn it into a plugin.
I’d love a multimodal AI that actually listens to a track or input and starts “jamming” in real time. Like a real band mate. Only, this one isn’t drunk half the time and constantly asking to borrow your car.
Man that music reminded me of Motown, so sweet 6:57
The lyrics are still very much ChatGPT, full of "so -adjectives-" and other ai-isms, though. I wonder, if there is something coming up that actually would write convincing lyrics?
I hear SO MANY ai-isms lately, people don't even bother touching up, just slap the first generation after a generic prompt 😢
To be fair, I think the people who generated the samples used ChatGPT. You CAN write your own lyrics.
But, I believe music is fundamentally about human communication. For that reason, AI music (however impressive it may become) is really less than noise. It appeals to our intellect, rather than our emotions.
It sounds amazing. Yes, but does it move us?
and how does it move us 2 weeks later when we listen to it? after the wow moment
Yes it does. We are there now...Take advantage of this while still can.
I've used both and preferred the SUNO results despite the better sound quality on UDIO.
i use Suno to make songs, then rewrite my own instruments to the vocals. I cant wait for competition! Anything to help me get better vocal quality will be great!
I've 100% done the same! It's so much fun! I get to have a vocalist without the ego!! (says the guitar player! Ha!)
@@TheoreticallyMedia oh dude Im a guitar player too, were OBVIOUSLY way cooler than those darn singers! lmao
In case you're interested to hear how various genres of music sound when produced to full tracks, check out: www.youtube.com/@NeverheardTunes
is this video endorsed directly by Udio or subsidiaries?
It is not. Never got paid a dime by them. Interestingly, after this video they did contact me for an NDA for a day-- hah, and then I did a follow up video. But, wasn't paid for that one either.
I always disclose if a video is sponsored. I know not all do, but I take it a point of pride.
@@TheoreticallyMedia Much appreciated (Difficult to know in a lot of cases!). Caught the video - and enjoyed it (also followed up, actually funny enough on Udio now). I appreciate the heads up, and having worked with stable diffusion for ages/stabililty/meta, I didn't realize it was out! Apologies for the rather brunt question - but appreciate you answering it. Now if only to bridge the promting gap, and do something with these generic Udio voices :) (seem to have a innate talent finding them through manual). I'll tune/subscribe, earned that much at least :)
@@TheoreticallyMedia I can definitely appreciate that, sorry for the straightforward question - but it can definitely be hard to tell. You do a great job on the video's, and I've enjoyed catching what ones I have. I've been moonlighting over with Noisee lately myself (along with Udio/Suno). If your up for it, should drop by sometime - I'm sure they'd love to have you. Big things in the work on that end for Video generation (I have a few beginner ones on my channel, but nothing compared to some of the wonders these guys churn out). Ha, and also not paid/promoted :). I'm not sure (haven't checked yet), if you've gotten a video with them, but might be worth a look. I appreciate the kind reply, and thank you for illuminating a bit on it - keep up the good work!
What was up with the dialogue audio during the interview that sounded really weird.
Oh, that was Adobe’s audio cleanup tool. It gets a little weird sometimes. But the alternative was banging pots and pans in the background. They were setting up a party and things were super loud in the background and apparently my wireless mic didn’t bother recording us close up!
A little glitchy, but better that than really obnoxious background audio!
Just like art, supply will exceed demand. Music will devalue as a profession even more. What happens to all the jobs that go around creating a product someone can create just in a paragraph of text
Well, I still say that a prompt can’t play a live show. That human connection of a musician or band, playing anywhere from a stadium to a local bar- that alone will keep music going just fine.
I've just started having a play with AI. As a learning tool, i think it's brilliant. I can now pull songs apart and write down the correct parts that were once impossible to distinguish by ear. As far as using it in a final production goes.....no way. You just become another sheep in the cut and paste society. If you physically can't play a part , you'd better get out of the kitchen, 'cos it's way too hot for you, or do the work and get it under control.
It’s impressive..sound Tribe called Quest
finally a proper AI audio.
Has anyone tried to register?
suno is the best for Disco song & old stuff & udio is good for techno , acid stuff more underground can be found but on suno it's always the same techno kik & udio ave to much artefact in it... for me Suno is the best for now cause i make the lyric but udio love to play real sample THAT BAD !! but suno only make the voice sing , no sample added .. that was my experience .. and you ? :)
Suno is like winning the Lottery ! 1000 prompts to get a decent song ! The sound is like 8 bit mixing !
The quality declined in the last days, i will try Udio.
WOW.! jORDAN RUDESS..! LOVE DT 🤟💌
I signed up and now they are asking me to contact you since you linked me there and to help expedite the processing! Can you help? :)
When is this coming out?
Not sure-- technically, we're not even supposed to know about it yet. But, the rumor I heard is that they'll surface on 4/20. That said, I'm not sure that wasn't a stoner making a joke. We'll see!
Impressive.
is there a one where it can use your voice as the artist?
Love it or hate it but AI music is gonna be the greatest thing for those that have the mind for creativity but not the talents to showcase it
@2:23 - It reminds ME of that 80s commercial with the various phone messages and one was "Wait for the beep, you gotta leave your name, you gotta leave your numba".
they might be making some disney sounding music - let's hear some ROCK. where's our Zeppelin.safetensors (lora) or our Helmet.safetensors (lora)
I think I KINDA remember that one! It's lodged back there in the memory banks somewhere for sure!
@@TheoreticallyMedia - No, you remember it - there's no KINDA...
If I use custom lyrics it speaks them every time, even with [harmony singers]. If I let Udio generate the lyrics it will sing them.
When is there going to be an AI that can take an original melody, and generate an arrangement for it in prompted styles?
If this service can create longer than 2 minute songs, they are already ahead in this competition.
Just think... Without all these AI music generators, we never would have have been gifted with obscure vinyl rarities such as "I Glued My Balls To My Butthole Again", and I'm just not sure I want to live in a world where that doesn't exist.
oh wow ok things are getting very exponential right now
That upward slope is going to feel like a rollercoaster for the rest of the year, I'm sure of it!
Second example contained some artefacts, even listening without headphones it was noticeable that in both examples drums had poor transients and releases, everything u saw in those examples are very typical things, like strings in oldschool soul for an instant. Suno does better in here I guess. I was recently weirdly surprised about a jazz song that Suno made , with quite complicated harmony and some really creative licks containing like modal interchange stuff. I was like how does ai understand how to use modal interchange in musical context?. After listening I realised ai just knows music theory based on analysing existing content and it can create crazy modulations etc, but it’s mostly like adding random chords or harmony tricks that actually should work in certain moment. Ai does not feel the rhythm yet, although it can create Melodies with quite expressive dynamics (at least Suno does). But tbh we are musicians we shouldn’t worry about computer learning to feel because it will never feel the way we do because… it is just a computer.
Suno is great if you have enough patience and credits.
Agreed. It is also a lot of fun, if you aren’t trying to generate something specific and just playing around for memes and whatnot, I think it’s pretty great.
But yeah, all of these are going to require more control for folks who are trying to use it for super specific projects
it sounds like Native tongue era for sure i think closes would be De la soul