Thank you for the great overview. I think the most important feature to request, for actual control, is for the AI to be able to modify previous generations using a follow-up prompt like "This, but make the bass less pronounced and make it more upbeat". Coming from stable diffusion background, the thing that made image generative AI actually become a good tool for artists was the ability to work iteratively - get something close-ish to what you want, then refine it more and more and more with each new attempt. If we have to start from scratch every time, we're never getting exactly what we wanted.
If we ever get a non-stunning week in AI, I'd be happy to just keep listening to you talk about AI Music. You both seem to know your stuff and not be too elitist about it, a rare combination.
I love how you don't edit your videos with a bunch of annoying camera cuts like everyone else. You content is always very solid too. You are the best in the AI space period!
Haha, Yeah-- I don't really do the super cut-y zoom in, whoosh sound thing. I mean, a large part is that I do my own editing, so...I'm lazy? But yeah-- that ultra YT style just isn't my thing as well. It's funny, my pace (if we aren't diving into a tutorial section where we're looking closely at something) is about 5 seconds per shot. Which, is still kind of fast. I just don't feel the need to do a bunch of bells and whistles in that time. That, and I always just trust that you guys all have an attention span longer than 10 seconds!
@@TheoreticallyMedia all the other youtubers have so much crap flying in front of the screen it gives me anxiety lol i know the reason most of them do it is because for a time it worked.. now its just out of hand
The problem is, if I don't want to listen to anyone else's AI generated music, nobody is going to want to listen to mine. I'm running into that post-pandemic a lot with TH-cam videos. Nobody cares any more about watching other people's content. We're all just creating a custom vanity mirror to look back at ourselves. That's how humanity dies...
Ngl: most people don’t want to listen to anyone’s music! Or, at least that’s what my own Spotify metrics tell me! Joke aside, you’re right. Already seeing that on various socials, flooded with other people’s Udio and Suno creations. And that’s ok- maybe this isn’t meant for outward facing “audience” music, but rather custom created individual song that you make for you. Like, “earbud music” not Aux connection music.
@@TheoreticallyMedia Yeah, I've used udio this week to turn old poems into songs, and give them an actual chance to travel and a new audience (nobody reads poems anymore, but being lazy they may listen to a song). If you were to set up a streaming radio station playing an AI song, then as that plays, creating the next one up in the que, and just broadcast a steady non-stop stream of AI created songs... would anybody listen to it? Yes, at first, it might be fascinating, but in no short time, if a listener had half a brain cell firing they'd think, this is all just autogenerated crrp to tug at my emotions. That being said, the supply of half of brain cells is diminishing in this burgeoning idiocracy.
I've been using AI just to help with my creativity and potential portfolio with regards to knowing how to use AI. Not really to monetize TH-cam channels.
Through watching this I am reminded how important it is to hear about these tools from actual musicians and producers, and how much less annoying it is when someone with cultivated music appreciation/discernment and taste is analyzing their output.
Just been spending the evening with Udio... all i can say is OMG. The depth... the virtuosity is incredible. I accidentally started a generation about being frustrated with a broken machine (was some default offering). It has now turned into an epic metal anthem about breaking free from societal oppression... with scorching music... a raw female singer... Incredible lyrics. (Udio AI writes amazingly. No lazy rhyming but really thoughtful and evocative prose) I have become totally obsessed with this track. Just trying to finish it off now. On 23 generations... but lots of that was experimenting and getting to know the app. Thanks for sharing the link... was having trouble finding it through Google. This app makes Suno v3 look kinda silly. Its not only my own song... but some of the ones Ive heard from others are breathtaking... this is next level stuff. I am totally impressed.
by the end of the year i bet we will have granular control over generations and if you know some music you could actually tell it to use certaing chords and melodies
So far it's been creating more interesting tracks for me than some of the well known alternative bands of the late 70s and early 80s not big hits but better than the fillers on a lot of those type of albums, imagine if they had access to this tech back then their minds would have been blown, it would have been alchemy.
They're SO stupid good. One of my favorite things is to try and take just a little section of a Tosin riff or solo and try to figure out how to cop it on a 6 string. Not at all note for note, but just his ideas on phrasing and rhythm. I can't keep up with trying to play like them though! Tried an 8-string once and got totally confused!
All that’s missing is the granular controls. I made a song so catchy, with my own lyrics from way back. Tbh the only reason I’m sad that the site has crashed is I want to listen to it again ☺️ - I’ve been a songwriter for 35 years.
I’m a session guitarist and I love this, anything that takes music away from big business is great, if you don’t want A I support live bands by going to more gigs.
Tim at 1,200 songs per month FREE during the beta, what kind of volume do you see amassing? If Udio gains a thousand subscribers in the first month, that could be over 1 million songs (realizing that a lot of it might be just short riffs). But pre AI text-to-music, how many songs or tracks were published or copyrighted in say 2019, pre-pandemic (by everyday musicians, composers and songwriters)? The numbers become mind boggling when you think there will be multiple music generating platforms. You really have a great TH-cam channel! Keep up the great work!
Warning for Udio users: PLEASE fully read their terms of service, and while it is a bit swiss cheese that doesnt' mean it's not a worthy app for some: If you are interested in distributing music you've made: DONT, the Terms are a bit "wishy washy" in that they forget you can't upload directly to spotify and apple without a distributor, and it's a "NON COMMERCIAL PERSONAL AND ATTRIBUTIVE" sort of liscence, which means: if you're profiting off your music, try something else - I had to read it myself because i'm CURIOUS about AI music, because i use loops and samples myself that aren't AI generated and hoboy: I'm not going to attempt to use Udio and have to pay extra for attribution, and then get stuff taken down because they claim it before I can do anything.
It wouldn't make any sense to begin with to publish the output as one's own and try to make money wiith it. it would be like playing the lottery. There would be nothing to actually identify with for the "artist" using such tools to produce music.
After spending a small amount of time messing around I got a good feel for what was possible and have made 4 full length songs already that sound really good. You can tell that it's AI if you properly listen and it lacks a little bit of consistency, e.g. the chorus might not always be exactly the same. Some features I'd love to see would be greater control over instrumentation etc, more consistency in key elements of the like the chorus, main melody, etc and I'd like the option to trim clips or allow a little more fuzziness between the different extensions. My only complaint is that vocals do still appear at times during sections that are supposed to be instrumental only. Usually it takes a couple of re-rolls to get what you're after.
I'm totally in love with Udio. Every single time I ask for a music it gives me a beautiful track. Suno in disgusting, everything sounds identical. The sound is good, but the AI don't understand what you ask for. The tracks Udio do sounds very unique.
Stoked that more music stuff is coming out so you can talk about it. I also loved the short discussion last week. If you did a podcast that got actual musicians on to talk about AI openly, that'd be super cool. I'm thinking there's a whole world of people in the music industry, from producers to session musicians to ad jingle writers to the yes, rock stars, who all feel lots of different ways about it and are probably both worried and excited. Would be pretty great content, even if it's just a one-off or occasional thing.
So what makes a human have a point outside of the basic biological continuation of the species? Seems we are heading rapidly toward a point where all that makes us human is valueless
Totally understand where you're coming from-- but I don't really see it here. Maybe Spotify will get flooded with a ton of AI Tracks, but Spotify has been screwing artists over for years now. The way I see it: no one wants to see a prompt perform live. They want to see people. AI Tools like this can only help songwriters craft songs (instrumentals for vocalists, or maybe arrangement ideas for bands)-- There are a ton of other use cases-- in general I find that musicians are the least fearful of these AI Tools. Mostly, because they've had to adapt to technology so much over the years.
@@TheoreticallyMedia I concur that individuals derive inherent pleasure from both watching and participating in sports. This joy is unaffected by technological advancements; for instance, one doesn't cease running merely because a race car can outpace them, nor does one stop swimming in the presence of boats. The essence of human nature lies in the desire to witness others excel in their performances. While AI may surpass humans in many tasks, possibly even in creating music shortly, this doesn't diminish its value. Instead, AI represents a distinct evolutionary path. I believe we will harness AI as a tool to augment our enjoyment of various activities. The unique experience of observing human performances is irreplaceable, neither by AI nor by robots
I just realized that AI will take over recorded music but perhaps there will be a market for bands to recreate the AI generated songs, live. For instance, it'd be super cool to hear a band produce that Zappa-like fisherman song on stage. Though folks may be more interested in watching holograms just mime the AI songs live.
For a while, but with robots comming next year i bet it wont take long before they take that spot aswell, not to mention the hologram that perform on stage.
@@davidshaw3374 P.Diddy would have loved this copy/paste back in his days. Can't believe how much credit he got for just rewriting lyrics to well established hooks. Flipping breaks and making things your own is cool, but can you imagine just taking the hook from "Smells like Teen Spirit" and just changing the lyrics and saying you wrote a hit song?
I am a Suno fanboy and have been using it since V1 and it has accelerated my songwriting 10x. I thought V3 was a game changer and that Suno was so far ahead of the competition. People were hyping Sonauto as the Suno killer which I thought was mid at best. Udio in BETA has just laid down the gauntlet!!! It sounds amazing. The website is getting crushed right now and generations take forever. The 30 second generations don't measure up to the Suno V3 2-minute generations. But this is freaking BETA. AI music is here and is only going to get better.
I've been using this app for a day, for instrumental rock/metal ideas it's pretty good. Most of the chorus progressions are kinda similar which I don't like, but its a similar formula. But it does have some cool melodies on guitar to get ideas from. 👍
One problem I've had with other AI music generators is when I try to sample a section of the output and then use Ableton Live's "convert harmony/melody to new MIDI track" feature. It seems like the bitrate is so low and there's so much artifacting that the MIDI extraction ends up being so extremely noisy and chaotic that it's unusable, and it's just quicker to listen and manually play the part. (To be fair, this does happen to a lesser degree even with very clean high quality source material.) But I'll be interested to experiment with that feature with Udio's output.
@@Edbrad I think so too! But I wonder if some of these services are "watermarking" their generations by omitting particular narrow bands of frequencies in both time and Hz.
I was against most ai but mainly the ones that stealing the likeness of other artists but for someone with thousands of random random songs and lyrics over the years and an ear for what I want to hear as production, this site has been the most exciting thing ive experienced in years. It helps all these lost lyrics find a home. I treat the creations like samples and flip them in creative ways and you can argue that it’s less destructive than sampling people’s music without permission.
I have been playing with Udio all day and have yet to generate 1 song that matches the quality of any of the Suno AI (v3) songs I have made. I am starting to wonder if all the hype was just for nothing. Everyone was on twitter saying that this is going to be the Sora of audio but I am failing to see it.
That said does not discount what Udio will be in the coming months or years. Really cool approach to music generation and nice to see more options in the market to be able to compete.
Honestly I couldn't agree more. The only thing that I noticed is on Suno the voice becomes robotic after some time where Udio sounds more like a real voice. Suno music is better imo for sure though
3 หลายเดือนก่อน
The song: "Echoes of the unknown silence" on Udio changed my mind of AI and music. Man ... a bit in when she really starts to sing..wow.
as a musician who plays many instruments, with jazz/classical/metal training, and produces in a daw, i'm excited by the prospect of collaborating with a.i. tools like UDIO. i already use AI tools like Splice Stacks and Captain Melody plugins, as well as Chat GPT for brainstorming. people who are afraid of being replaced are just being alarmist
You’re speaking my language here. I hop around a lot of genres as well- and while my theory game is so/so, I often rely on Scaler (which I consider “dumb AI”) and ChatGPT for assists. Splice, as much as I’ve used them in the past, really has it coming. They’ve been pretty terrible in terms of optimizing their search, and DAW integration has been a mess. I’m really looking forward to the day where I can generate a kick/snare, or a keyboard pad like you can with image elements in Photoshop Gen Fill. I know people will complain that “the AI is doing the work” but many of those people who say things like that are the ones who aren’t actually doing the work. I see all of this as a fun and interesting way to create and learn more. Glad you commented! It’s always nice to hear from others on the same wavelength! (Pun intended!)
Alarmist? Are you delusional or maybe still in denial phase? AI will replace most of the musicians in whole world. And it will do that very, very quicky. Especially Electronic music artists. Nobody cares if song was made by AI or by human as long as the song is good. AI just does it billions times faster and with little to no effort. There is no need anymore to learn music theory or learn difficult music programs as ANYONE can now make perfect music in few minits
OMG, this stuff is so crazy. I mean i am making music by myself, it is mostly for fun. But this Ai. I mean a whole son in about 1 hour. I need at least 2-3 days, and thats only if i have an idea. WOW. The Future is AI, for sure.
I'm impressed by how AI evolution keeps getting better, improving tools. While udio has its flaws, being in beta, I believe it will evolve even further over time. I even tried creating my own music, and I can say it has been an interesting experience.
I'm excited to try Udio and it does seem to have some innovations on Suno (I love the ability to add an extension before and after the current generation). Personally, I wouldn't listen to any of these tracks and there are Suno tracks I would listen to so I'm not sure it's overall really blowing me away but we need more time to see what it's capable of.
from the terms: "You acknowledge and agree that we may use your INPUT Content and Output to train or otherwise improve our Services, including our artificial intelligence and machine learnings model(s) related to the Services." "The Services may allow you to download copies of another user’s Output, and similarly other users may download copies of your Output." So the songs are not private here...
They are not. I'll circle back to them in a bit, but from talking to them-- apparently in terms of the licence, you (the generator) have it. They keep a clause in the licensing to be able to use the output for ad purposes, but that's it. Copyright is up in the air-- but that's an overall AI issue. If you generate an instrumental track and then sing over it, obviously that's yours and won't appear in any Ad material. Again, at some point when things calm down (it is launch day) I'll talk to them again and get answers for all this stuff. In the meantime, just have fun generating some songs! (If you can get it to work today!)
@@TheoreticallyMedia Yea I just went through the terms and they do mention a private mode. But I couldn't find any privacy function on the interface. So as of now, sadly, with no private mode available I can't touch it.
Wow, I've discovered covers of existing songs and melodies there, probably from one of the developers, because I haven't been able to discover this feature anywhere yet. So this is already technically possible! I hope this will soon be available as a feature, but then of course it will quickly become difficult from a copyright point of view. So we will have to hope for the rapid appearance of an open source alternative!
It's funny, I kind of slag on it in the video, but it is also the one song from the video that keeps getting stuck in my head...so...there is something to it!
So I've been experimenting with Udio for the past day and......... I'm quiet impressed! 👍 👍 👍 The only glitch is when I got a warning about my custom lyrics being forbidden ( I had ChatGPT write me lyrics for a song about the troubling politics in the US in 2024).
@@flyversusfly76what’s the problem, I’m old enough to remember the start of MTV, video killed the radio star, Also people kicked off about the internet, MP3s even TH-cam, personally I think this is brilliant and can only get better, I’m a session guitarist, I can now make the music and video I want and do it all at home, just look at the music scene over the last ten or more years, A I is definitely a good thing.
@@hardlines2635 it‘s cool that you see the potential. Me personally I think why should anyone pay for a session guitarist when I can Click some buttons to instantly have 10 great guitar riffs „in the style of“ fitted right my Song? I think there will be a niche for human music as an Art form still.
It’s depressing that AI is allowing people to cut corners like this instead of actually learning the skills required, all while completely devaluing real musicians who have spent years mastering their craft
It occurs to me that creative enterprises are becoming less like noodling around for inspiration, and more like shopping where you are looking at a bunch of stuff and selecting what resonates with you.
I don't know if it's your fault for sending so many people over, but it just keeps saying there is an error and to try again later. "Generation is under maintenance" . I will give another try later. They probably need to update their system to handle the traffic. This is my favorite AI channel these days. Thanks for keeping us in the loop!
That's wild and... the end.of platforms who offer premium "royalty free" mediocre music for adverts etc... for non musicians. But that's life. The Zappa part was just crazy...
Oh yeah-- If a company like Splice isn't already working on something like this, then it'll pretty much spell their doom. Come to think of it, I should probably cancel my Splice Subscription...
So far Udio seems interesting. Still a long way to go though but for sure a step in the right direction. I like it so far. Competition breeds excellence. :D
OK, now that’s more like it! It just might be better than Suno! Guitars sound better than Suno and I’ll be putting it through the ringer over the next few days. Thanks for the quick update from yesterday. Once a paid plan is ready, pricing won’t matter as far as I’m concerned bcuz this is amazing stuff, exciting times!
As a long time guitarist-- the guitar sounds are INSANE. Not just the shred stuff either, I killed the "Bon Jovi Blues" song, but it had a pretty rippin' solo toward the end. Like, I was honestly thinking: I want to learn that solo!
@@TheoreticallyMedia - Same here. That was enough to convince me to sign up immediately. And after hearing about stems and comparing the GUI with its competitors’… I died and went to heaven.
After playing with it a little bit, seems like Udio has better overall sound quality, but Suno has much better composition. They'll each get better at both in time with user reinforced training though.
From what I've heard "in there" (I can't test it at the moment, the website is under maintenance), IN MY OPINION, sonically, in terms of clarity and complexity, it is clearly SUPERIOR to Suno.
Thanks Tim! Great video and nice to see that your music taste is great (Dream Theater, Zappa etc. Wow! Just my taste as well). Really enjoy your humorous way of thinking and making videos. btw we hae some commonality in comics as well as I´ve done them too (found on my channel as a youtube video). Thank You!
You really know a hell lot about music. I am very impressed with Udio. It's def not there yet but it's approaching at what seems like light speed. Very curious where they will be say half a year from now.
totally understand-- but in my mind: Because no one will ever pay to go see a prompt play live. This is going to disrupt Stock, Type-Beat, and the Loop producer arenas...but, other than that, I just see this as a win for musicians. If you have an idea, you can sketch it out with something like this. Or just generate drums only for a drum loop-- or (eventually) feed is a half finished song to see what it makes. It might be good, it might suck-- but, it'll get your head rolling. It isn't going to replace us. No machine can survive the amount of beer that gets spilled on me during a gig.
Comments that are critical of Udio, such as @CypiXmusic and @javier_afonso (the latter of which was really eloquent and to which the channel creator offered no response), are actually being removed. Shocking, right? Anyway... Artists, from here on out, let’s go on a modified data strike that should make everyone, and I do mean everyone, happy. CREATE NOTHING NEW. Until we are adequately protected, let’s take remix culture to its most extreme ends. The pro-AI “art” community has told us repeatedly that there is no such thing as originality, everything is derivative, etc. They clearly believe we’ve reached the limits of human creativity and somehow feel qualified to tell us that art has become “stagnant.” (Never mind that these art whisperers include the same tech moguls who have fed the masses mediocre art through algorithms for decades.) Let’s not fight this strange alliance between billionaire tech moguls who are somehow suddenly democratizers and everyday people who believe themselves to be decent (or even progressive) but have no qualms about stealing from artists, including the majority of whom are poor or middle class. Let’s instead carry out their self-serving logic to its ridiculous end. Let’s turn the other cheek by doing what they’ve told us to do so many times: Be closer to their definition of “creative” and “adapt” as they have. Until artists are adequately protected and AI companies and grifters cannot scrape copyrighted works without artists’ consent and without crediting and fairly compensating each artist, stop feeding the machine. Use your advantage, which is skill. Make friends with and collaborate with other human artists who are against us being exploited to train generative AI. (This may be the time when you find out who your real friends are. Collective suffering brings oppressed people together. Unite.) Create remixes of each others’ works. Remix the remixes, too. And so on. These will be protected by copyright while AI outputs are not. Your skill will enable you to do this somewhat efficiently, and you will be less attached to these creations than you would be to the work you “dared” to call original. That said, never stop working on art that you truly love even if you don’t release it or only share it in special circumstances where the cost-benefit analysis makes sense. If you like creating original art, this kind of data strike means you may have to do some work you do not love (derivative work) but it will still be art (you will still probably like and enjoy it) and not a day job you hate. If you want to use AI (as you are constantly told you should) to ensure that you stay on top of the tech or to reduce the amount of original, human work available for corporations to scrape (and thereby contribute to the logical consequences of AI feeding itself its own ****), there are ethical models. (See Fairly Trained.) Also, keep pressuring your elected representatives to end to this madness.
UPDATE: Udio has already fixed the issue with consistent choruses that I mentioned below. It can now understand what's in your track within a 2 minute window whereas before is was only a 30 second window. That's impressively quick for support. Now, for me, this is now beating Suno, especially since Suno has run into issues with version 3. I've been able to easily extend Udio clips to over 3 minutes. At this point, I'm just waiting for the musician tools and an increase in clip length. Stems, or hopefully every track separate, will be prefect and that seems to be on Udio's radar. *** Having used this for a few days, there are definitely some things missing I'm sure others who write songs with lyrics may have noticed. I wasn't able to get consistency. Not even close. The extend feature often results in chaos. trying to get consistent verses and choruses isn't easily achieved. What I would prefer would be to see an app where you get to work on single sections without having to worry about fitting things into a 33 second window or some other number. Generating things and crossing your fingers isn't working for me. For example, it would be perfect if we had an option to work on the intro, verses, pre-choruses, choruses, bridge and any desired instrumental sections separately. Once and initial selection for a section is made, they should remain consistent with a scalable amount of variation in the music and melody via perhaps a slider that determines the amount of variation desired. Once worked out, we should be able to chain those sections as desired within an "arranger" that provides smooth transitions between the sections. And if we don't like that arrangement, we should be able to change it by re-ordering the sections and have the app automatically fix the transitions between sections to match. That would be much closer to how many songwriters work things out. Maybe this is asking for too much in the near future, but if this could be considered, I think everyone would be happy. For now, I'm doing a lot of offline editing and a lot of generating that wastes credits.
After reading the positive comments about this, I thought I was the only one that thought Suno sounded better. Yeah something about the voices on Suno being more crisp
Udio sounds way better to my ears, a lot cleaner than Suno... actually I thought the difference was drastic on first impression but I'll bring them into ableton and a/b them.
can you use a made and downloaded song in a youtube video? or is that not allowed with udio beta? or burning your mp3 you made with udio and burn it to a cd and share it with friends?
The AI's Jacques Cousteau reference at 9:53 is probably from Wu Tang Clan - Da Mystery of Chessboxin. There is a lyric with a similar reference by Ol Dirty Bastard in that record.
So what I get from UDIO is that it's to Jukebox, what ElevenLabs is to Uberduck; Like it'll still make music with better audible fidelity and consistency than Jukebox, but it can't generate output from pre-existing materials/attempt to copy one artist's specific styles (like that one Bruno Mars sample from the Jukebox library)
@TheoreticallyMedia Good stuff as always, Tim. If you had asked me 3-4 months ago how long it would take to get to this level, I would have guessed at least 5 years. It's amazing to think that THIS is where we are today, and it will only improve. Of course, how we, as musicians, ultimately utilize the tools will be the real challenge.
That's a VERY good point! Udio is just in 1.0 Beta, so give them some time. I'm not a "this vs this" kind of guy, so I'm just happy we've got two really good music generators!
Oh, that's just some stuff I put in so it wasn't just a static screen. Random videos, Animals As Leaders, and a band called Turnstile for the punk rock section.
All the people who worked so hard to make music over so many years, now some tech bros with no talent can profiteer off of scraping and copy-pasting their work in an icky digital mish-mash. Yippeee! Definitely their mothers can be proud, as they finally solved the problem of the average musician just being too rich, and the average tech bro not rich enough.
Artists, from here on out, let’s go on a modified data strike that should make everyone, and I do mean everyone, happy. CREATE NOTHING NEW. Until we are adequately protected, let’s take remix culture to its most extreme ends. The pro-AI “art” community has told us repeatedly that there is no such thing as originality, everything is derivative, etc. They clearly believe we’ve reached the limits of human creativity and somehow feel qualified to tell us that art has become “stagnant.” (Never mind that these art whisperers include the same tech moguls who have fed the masses mediocre art through algorithms for decades.) Let’s not fight this strange alliance between billionaire tech moguls who are somehow suddenly democratizers and everyday people who believe themselves to be decent (or even progressive) but have no qualms about stealing from artists, including the majority of whom are poor or middle class. Let’s instead carry out their self-serving logic to its ridiculous end. Let’s turn the other cheek by doing what they’ve told us to do so many times: Be closer to their definition of “creative” and “adapt” as they have. Until artists are adequately protected and AI companies and grifters cannot scrape copyrighted works without artists’ consent and without crediting and fairly compensating each artist, stop feeding the machine. Use your advantage, which is skill. Make friends with and collaborate with other human artists who are against us being exploited to train generative AI. (This may be the time when you find out who your real friends are. Collective suffering brings oppressed people together. Unite.) Create remixes of each others’ works. Remix the remixes, too. And so on. These will be protected by copyright while AI outputs are not. Your skill will enable you to do this somewhat efficiently, and you will be less attached to these creations than you would be to the work you “dared” to call original. That said, never stop working on art that you truly love even if you don’t release it or only share it in special circumstances where the cost-benefit analysis makes sense. If you like creating original art, this kind of data strike means you may have to do some work you do not love (derivative work) but it will still be art (you will still probably like and enjoy it) and not a day job you hate. If you want to use AI (as you are constantly told you should) to ensure that you stay on top of the tech or to reduce the amount of original, human work available for corporations to scrape (and thereby contribute to the logical consequences of AI feeding itself its own ****), there are ethical models. (See Fairly Trained.) Also, keep pressuring your elected representatives to end to this madness.
i tested this, I'm just amazed, the flexibility, being part of Theoretically Media 3X your capability with AI. Such a great commentator and brings great Value with every video.
I thought Suno Ai was rad... I mean, V3 is still great but ound? Holy molly maceroni!!! That's another level! And the control is also more advanced than with Suno! ... Crazy. I'm really speechless... I think here too, as with every AI, you're bound to come across a repeating pattern at some point, but still! Just the fact that mam can combine completely absurd style directions with each other relatively easily and without cramping AND that it still sounds good or at least interesting... Let's see if I can still get my metal song realized the way I want it to be :)
Metropolis Part 3: publish it. Would love to hear the rest of that one. I am wondering where to go from here? I am thinking: gen tracks. Slice and dice into segments/stems and then onto DAW for composing. Thoughts?
So, Metro is actually on the "Trending" page-- if you can get in (ha, I know the servers are slagged right now), you should be able to search for it. It is GOOD. I only played that clip, but I really had a hard time deciding where to clip it! Also: Stems are coming from Udio! Not at launch, but it is in their roadmap!
You sir are always on the bleeding edge. Thank you! My Udio generations are taking a long time... what were your average generation times? (watched your whole video, still waiting for my 1st gens to complete, haha) Update: 1st 2 generations took 20 minutes. Next 2 took 33 seconds! 🤯
Sorry to bug you again, but you by far explain things better than others so you are stuck with me :) Two questions (1) I don' t get the work flow....When ready to create the song, I click the CREATE BUTTON --- when udio is done creating the song, next I clicked the EXTEND BUTTON. I'm satisfied with the song, so I'm done and ready to do a new song. BUT HOW DO I GET BACK TO THE CREATE BUTTON? I seem to be stuck on the EXTEND BUTTON and could not figure out how to get the CREATE BUTTON back. Instead, I exit Udio and open it again to get the CREATE BUTTON. (2) second question. Is there a way to pick the singer you like. I love the female singer I got on my first song and want to use her in other songs but keep getting singers I don't like. Thanks for your help. I love your videos but as a retired vet on a fixed income I'm always counting pennies.
The "Zappa" section has an interesting bit. I call it Zappa, but if you know your prog, I also hear Yes in the vocals! I didn't mention it in the video, or I'd end up spending 50 minutes rambling about prog, and then someone in a cape would show up to do a keyboard solo. hah, no one wants that!
@TheoreticallyMedia 😆 🤣 😂 Yeah, that Zappa-ish song is interesting indeed. It shows that Udio can produce listenable off-tempo weird compositions too. That seems even more challenging...... but I'm not a musician so I could be very wrong on this.
sure, i can have an AI to compose songs for me, another one to mix, and one to master. then i could spend my days on the beach with all that free time! lol
in the future you will probably generate one single song in your prefered style, and then have an AI agent create entire albums and a virtual band based on it with music videos included
Perhaps theyll sample your voice and create bands that you can own the rights to basically based upon your description of structure and aounds, hord progressions, instruments, etc then your band has a sound and style created with an albums worth of songs ready to practice and go on tour with. Maybe even sample your guitar and if youre sloppy or slow they could fix it up or make something similar as mentioned before... etc
I have a question? I am a paid subscriber with suno. With that all content I create is owned by me and can be sold commercially. Currently Udio is free. Does that mean whatever I create they own the rights too? If so I would rather wait til they have a subscription plan similar to suno.
So, should I stop my chord studies? I mean, what is the point? A life without effort is pointless. You know someone's going to try and sell these songs and then the lawyers are going to drop on this like a ton of bricks.
Yeah, like you mentioned at the end. If they really want to help artists to make something cool, there need to be features like: prompting the BPM of the song, the time signature, the key of the song, maybe even the chord sequence and of course stems and maximum control over adjusting certain parts (maybe even per bar). A system of adjusting the details is necessary to co-create with A.I. Sound wise it seems a bit better than Suno V3, but the editing features are too limited. The devil is in the details... I want control. That's where you can raise the quality of a track. I wonder how long until I can upload a rough demo, make it into a finished song and then when I'm happy prompt: "Mix it like Chris Lord-Alge."
Your comment made me laugh because I have one of the Chris Lord Alge plugins and no matter what I do-- it isn't coming out like that!! Yeah, all of that is forthcoming. I'm following up w/ them on everything that's being brought up in the comments. I'll be back with another video at some point-- a deeper tutorial dive and (hopefully) some more answers from them!
@@TheoreticallyMedia At this point not only spootify must be worried the Big DAW dogs (in the likes of Cubase) as well ,I can certainly imagine future versions not only take chords, melodies but you will be able to do sequencing, add virtual instruments and the whole production by just prompting ,full control is definitely every songwriter dream when it comes to AI generating music
I'm trying to generate a country song ( I tried with a lengthy description and with a short simple description), but it keeps saying there was an 'error'. Not sure what to do. Did that happen to you?
Oh, I love Suno too. And they’re about to drop a pretty major update. Suno and Udio are both arm wrestling for best music Ai, and at the end of the day: we all win!
I find this pretty interesting. As a musician and recording artist, this is largely irrelevant to my work- I am not going to stop doing my creative work simply because someone can press a button and make a song. Where it does become intriguing is as a license-free source of samples to chop up and work into my original compositions. I’ve always been bothered by the idea of taking other people’s music and claiming it for myself as a sample. Potential issues with training data aside, this feels like a more ethical way to produce unique and royalty-free samples.
Yup! You got it right there! I've been doing a lot of the same since like...well, pre-Suno. Back when it was still RiffDiffusion! And hypothetically, you could generate just drum loops with something like this. Or just piano chords. One thing for sure: If Splice doesn't get it together, they won't be around much longer.
Nothing that is this easy has value. Never had never will have. It means music will loose its magic and value totally. If you cant see that you where born yesterday.
Hi, just tried it - really great sound quality. Would you post a feature request for me (can’t find a way to do it myself)/? I really wish I could post a melody (as midi or just written) and have it arrange a song in a certain style with this melody. As amateur musician this would make it instantly more useful and a learning tool at the same time.
Very cool, the sounds are great and I was literally talking to my wife about this topic last night. Currently the generation isn't working for me (something about being under maintenance) but listening to the music already there is nice.
Big update to Udio! Check out the details here: th-cam.com/video/W471Vjeyhjo/w-d-xo.html
Thank you for the great overview. I think the most important feature to request, for actual control, is for the AI to be able to modify previous generations using a follow-up prompt like "This, but make the bass less pronounced and make it more upbeat". Coming from stable diffusion background, the thing that made image generative AI actually become a good tool for artists was the ability to work iteratively - get something close-ish to what you want, then refine it more and more and more with each new attempt. If we have to start from scratch every time, we're never getting exactly what we wanted.
If we ever get a non-stunning week in AI, I'd be happy to just keep listening to you talk about AI Music. You both seem to know your stuff and not be too elitist about it, a rare combination.
I love how you don't edit your videos with a bunch of annoying camera cuts like everyone else. You content is always very solid too. You are the best in the AI space period!
Haha, Yeah-- I don't really do the super cut-y zoom in, whoosh sound thing. I mean, a large part is that I do my own editing, so...I'm lazy?
But yeah-- that ultra YT style just isn't my thing as well.
It's funny, my pace (if we aren't diving into a tutorial section where we're looking closely at something) is about 5 seconds per shot. Which, is still kind of fast. I just don't feel the need to do a bunch of bells and whistles in that time.
That, and I always just trust that you guys all have an attention span longer than 10 seconds!
@@TheoreticallyMedia all the other youtubers have so much crap flying in front of the screen it gives me anxiety lol i know the reason most of them do it is because for a time it worked.. now its just out of hand
Live performances and original human creationswill just be all that more precious and respected. A rebirth.
Yup. That’s my take. Spotify has been horrible for musicians. If it crashes from all this, I have no problem with it.
i can actually imagine a band that plays this ai generated songs live
@@robotron07Absolutely possible 😅
oh great bad news for Hikikomori 😂
No, live has been ruined for decades now.
The problem is, if I don't want to listen to anyone else's AI generated music, nobody is going to want to listen to mine. I'm running into that post-pandemic a lot with TH-cam videos. Nobody cares any more about watching other people's content. We're all just creating a custom vanity mirror to look back at ourselves. That's how humanity dies...
Ngl: most people don’t want to listen to anyone’s music! Or, at least that’s what my own Spotify metrics tell me!
Joke aside, you’re right. Already seeing that on various socials, flooded with other people’s Udio and Suno creations. And that’s ok- maybe this isn’t meant for outward facing “audience” music, but rather custom created individual song that you make for you. Like, “earbud music” not Aux connection music.
@@TheoreticallyMedia Yeah, I've used udio this week to turn old poems into songs, and give them an actual chance to travel and a new audience (nobody reads poems anymore, but being lazy they may listen to a song).
If you were to set up a streaming radio station playing an AI song, then as that plays, creating the next one up in the que, and just broadcast a steady non-stop stream of AI created songs... would anybody listen to it?
Yes, at first, it might be fascinating, but in no short time, if a listener had half a brain cell firing they'd think, this is all just autogenerated crrp to tug at my emotions. That being said, the supply of half of brain cells is diminishing in this burgeoning idiocracy.
I've been using AI just to help with my creativity and potential portfolio with regards to knowing how to use AI. Not really to monetize TH-cam channels.
Through watching this I am reminded how important it is to hear about these tools from actual musicians and producers, and how much less annoying it is when someone with cultivated music appreciation/discernment and taste is analyzing their output.
Just been spending the evening with Udio... all i can say is OMG. The depth... the virtuosity is incredible. I accidentally started a generation about being frustrated with a broken machine (was some default offering). It has now turned into an epic metal anthem about breaking free from societal oppression... with scorching music... a raw female singer... Incredible lyrics. (Udio AI writes amazingly. No lazy rhyming but really thoughtful and evocative prose) I have become totally obsessed with this track. Just trying to finish it off now. On 23 generations... but lots of that was experimenting and getting to know the app. Thanks for sharing the link... was having trouble finding it through Google. This app makes Suno v3 look kinda silly. Its not only my own song... but some of the ones Ive heard from others are breathtaking... this is next level stuff. I am totally impressed.
by the end of the year i bet we will have granular control over generations and if you know some music you could actually tell it to use certaing chords and melodies
I'm totally obsessed myself here. And I'm just scratching the surface. Entering my own lyrics and doing songs for friends.
@@Ibuyoldbeer
Why bother write even own lyrics? Rather just let AI do it all. And your friends can make AI songs by themselves in one minut.
So far it's been creating more interesting tracks for me than some of the well known alternative bands of the late 70s and early 80s not big hits but better than the fillers on a lot of those type of albums, imagine if they had access to this tech back then their minds would have been blown, it would have been alchemy.
Your musical refs blows me away, I've been addicted to Animals as Leaders since I've discovered them 5 years ago
They're SO stupid good. One of my favorite things is to try and take just a little section of a Tosin riff or solo and try to figure out how to cop it on a 6 string. Not at all note for note, but just his ideas on phrasing and rhythm.
I can't keep up with trying to play like them though! Tried an 8-string once and got totally confused!
not only that but Primus as well! What a great taste
I knew this day would come but didn't think it'd be that fast! And it's v1.0 BETA! This instantly puts out most music production out of business.
All that’s missing is the granular controls. I made a song so catchy, with my own lyrics from way back. Tbh the only reason I’m sad that the site has crashed is I want to listen to it again ☺️ - I’ve been a songwriter for 35 years.
@@T1tusCr0w yeah its beta right now the server issue will dissapear over time
@@T1tusCr0wyou just wrote a prompt
@@johndoe_1984 🤣
@@johndoe_1984
So? It is all that is needed now. The era of human song writing is over.
I’m a session guitarist and I love this, anything that takes music away from big business is great, if you don’t want A I support live bands by going to more gigs.
Except the AI algorithms are owned by big corporations anyway.
My guy came through again with another Al music update banger. Thanks for all you do fam!
Tim at 1,200 songs per month FREE during the beta, what kind of volume do you see amassing? If Udio gains a thousand subscribers in the first month, that could be over 1 million songs (realizing that a lot of it might be just short riffs). But pre AI text-to-music, how many songs or tracks were published or copyrighted in say 2019, pre-pandemic (by everyday musicians, composers and songwriters)? The numbers become mind boggling when you think there will be multiple music generating platforms. You really have a great TH-cam channel! Keep up the great work!
Warning for Udio users: PLEASE fully read their terms of service, and while it is a bit swiss cheese that doesnt' mean it's not a worthy app for some: If you are interested in distributing music you've made: DONT, the Terms are a bit "wishy washy" in that they forget you can't upload directly to spotify and apple without a distributor, and it's a "NON COMMERCIAL PERSONAL AND ATTRIBUTIVE" sort of liscence, which means: if you're profiting off your music, try something else - I had to read it myself because i'm CURIOUS about AI music, because i use loops and samples myself that aren't AI generated and hoboy: I'm not going to attempt to use Udio and have to pay extra for attribution, and then get stuff taken down because they claim it before I can do anything.
It wouldn't make any sense to begin with to publish the output as one's own and try to make money wiith it. it would be like playing the lottery. There would be nothing to actually identify with for the "artist" using such tools to produce music.
Thanks. Exactly what I need to know about Udio.
@@dkpianistdudes won’t care and will just massively spam spotify. This is the end
Doesn't this mean you use it for initial versions and then just record over it for the originality?
Wait so does this mean I can’t have udio make a sample, I sample it and make a beat with it then sale it as a type beat?
You just made my day! And I can now log into Udio's platform.
After spending a small amount of time messing around I got a good feel for what was possible and have made 4 full length songs already that sound really good. You can tell that it's AI if you properly listen and it lacks a little bit of consistency, e.g. the chorus might not always be exactly the same. Some features I'd love to see would be greater control over instrumentation etc, more consistency in key elements of the like the chorus, main melody, etc and I'd like the option to trim clips or allow a little more fuzziness between the different extensions. My only complaint is that vocals do still appear at times during sections that are supposed to be instrumental only. Usually it takes a couple of re-rolls to get what you're after.
is the site safe cuz i got mcafee warning to not visit it I really want to try it though
Udio throw me an error whenever I try .. guess you sent a billion people up there :)
I'm sure they're getting HAMMERED right now-- give it a few hours to see if the servers cool down a bit!
@@TheoreticallyMediait’s broken right now because to much traffic and not enough gpu power for the server
This is already sounding way better than Suno, so it does make a lot of sense that it's crashing 😂
What's the website?!
@@NicVandEmZ wonder why :)
I'm totally in love with Udio. Every single time I ask for a music it gives me a beautiful track.
Suno in disgusting, everything sounds identical. The sound is good, but the AI don't understand what you ask for.
The tracks Udio do sounds very unique.
Stoked that more music stuff is coming out so you can talk about it.
I also loved the short discussion last week. If you did a podcast that got actual musicians on to talk about AI openly, that'd be super cool. I'm thinking there's a whole world of people in the music industry, from producers to session musicians to ad jingle writers to the yes, rock stars, who all feel lots of different ways about it and are probably both worried and excited. Would be pretty great content, even if it's just a one-off or occasional thing.
So what makes a human have a point outside of the basic biological continuation of the species? Seems we are heading rapidly toward a point where all that makes us human is valueless
Totally understand where you're coming from-- but I don't really see it here. Maybe Spotify will get flooded with a ton of AI Tracks, but Spotify has been screwing artists over for years now.
The way I see it: no one wants to see a prompt perform live. They want to see people. AI Tools like this can only help songwriters craft songs (instrumentals for vocalists, or maybe arrangement ideas for bands)--
There are a ton of other use cases-- in general I find that musicians are the least fearful of these AI Tools. Mostly, because they've had to adapt to technology so much over the years.
@@TheoreticallyMedia I concur that individuals derive inherent pleasure from both watching and participating in sports. This joy is unaffected by technological advancements; for instance, one doesn't cease running merely because a race car can outpace them, nor does one stop swimming in the presence of boats. The essence of human nature lies in the desire to witness others excel in their performances. While AI may surpass humans in many tasks, possibly even in creating music shortly, this doesn't diminish its value. Instead, AI represents a distinct evolutionary path. I believe we will harness AI as a tool to augment our enjoyment of various activities. The unique experience of observing human performances is irreplaceable, neither by AI nor by robots
You are now officially my helper for first suggestions for my next moves in the world of artificial song production. Congratulations! ;)
I just realized that AI will take over recorded music but perhaps there will be a market for bands to recreate the AI generated songs, live. For instance, it'd be super cool to hear a band produce that Zappa-like fisherman song on stage. Though folks may be more interested in watching holograms just mime the AI songs live.
For a while, but with robots comming next year i bet it wont take long before they take that spot aswell, not to mention the hologram that perform on stage.
Wait til there are celebrity robots, courtesy of Tesla and Elon Musk.
Yeah but they would have to pay the person who put in the prompt 😂 crazy
Still - no development is possible with these models. Just copying existing styles. Perhaps only a question of time until that hurdle is no more.
@@davidshaw3374 P.Diddy would have loved this copy/paste back in his days. Can't believe how much credit he got for just rewriting lyrics to well established hooks. Flipping breaks and making things your own is cool, but can you imagine just taking the hook from "Smells like Teen Spirit" and just changing the lyrics and saying you wrote a hit song?
I am a Suno fanboy and have been using it since V1 and it has accelerated my songwriting 10x. I thought V3 was a game changer and that Suno was so far ahead of the competition. People were hyping Sonauto as the Suno killer which I thought was mid at best. Udio in BETA has just laid down the gauntlet!!! It sounds amazing. The website is getting crushed right now and generations take forever. The 30 second generations don't measure up to the Suno V3 2-minute generations. But this is freaking BETA. AI music is here and is only going to get better.
First radio play of a Udio song was last night on Oystermouth Radio at just after 9:15pm. The show should be repeated tonight at the same time.
oh, that's fascinating! I guess someone had to be the first!
I've been using this app for a day, for instrumental rock/metal ideas it's pretty good. Most of the chorus progressions are kinda similar which I don't like, but its a similar formula. But it does have some cool melodies on guitar to get ideas from. 👍
One problem I've had with other AI music generators is when I try to sample a section of the output and then use Ableton Live's "convert harmony/melody to new MIDI track" feature. It seems like the bitrate is so low and there's so much artifacting that the MIDI extraction ends up being so extremely noisy and chaotic that it's unusable, and it's just quicker to listen and manually play the part. (To be fair, this does happen to a lesser degree even with very clean high quality source material.) But I'll be interested to experiment with that feature with Udio's output.
Udio sounds incredibly clean compared to that other one Suno
We need a Magnific for sound and music upscaling. Then it won’t matter the quality of the original.
@@Edbrad I think so too! But I wonder if some of these services are "watermarking" their generations by omitting particular narrow bands of frequencies in both time and Hz.
I was against most ai but mainly the ones that stealing the likeness of other artists but for someone with thousands of random random songs and lyrics over the years and an ear for what I want to hear as production, this site has been the most exciting thing ive experienced in years. It helps all these lost lyrics find a home. I treat the creations like samples and flip them in creative ways and you can argue that it’s less destructive than sampling
people’s music without permission.
Superb coverage, Tim. Thank you. Holy moly, this is what it must have felt like when they unveiled the wheel or the printing press.
I have been playing with Udio all day and have yet to generate 1 song that matches the quality of any of the Suno AI (v3) songs I have made. I am starting to wonder if all the hype was just for nothing. Everyone was on twitter saying that this is going to be the Sora of audio but I am failing to see it.
That said does not discount what Udio will be in the coming months or years. Really cool approach to music generation and nice to see more options in the market to be able to compete.
Honestly I couldn't agree more. The only thing that I noticed is on Suno the voice becomes robotic after some time where Udio sounds more like a real voice. Suno music is better imo for sure though
The song: "Echoes of the unknown silence" on Udio changed my mind of AI and music. Man ... a bit in when she really starts to sing..wow.
as a musician who plays many instruments, with jazz/classical/metal training, and produces in a daw, i'm excited by the prospect of collaborating with a.i. tools like UDIO. i already use AI tools like Splice Stacks and Captain Melody plugins, as well as Chat GPT for brainstorming. people who are afraid of being replaced are just being alarmist
You’re speaking my language here. I hop around a lot of genres as well- and while my theory game is so/so, I often rely on Scaler (which I consider “dumb AI”) and ChatGPT for assists.
Splice, as much as I’ve used them in the past, really has it coming. They’ve been pretty terrible in terms of optimizing their search, and DAW integration has been a mess. I’m really looking forward to the day where I can generate a kick/snare, or a keyboard pad like you can with image elements in Photoshop Gen Fill.
I know people will complain that “the AI is doing the work” but many of those people who say things like that are the ones who aren’t actually doing the work.
I see all of this as a fun and interesting way to create and learn more. Glad you commented! It’s always nice to hear from others on the same wavelength! (Pun intended!)
Alarmist? Are you delusional or maybe still in denial phase? AI will replace most of the musicians in whole world. And it will do that very, very quicky. Especially Electronic music artists.
Nobody cares if song was made by AI or by human as long as the song is good. AI just does it billions times faster and with little to no effort. There is no need anymore to learn music theory or learn difficult music programs as ANYONE can now make perfect music in few minits
Yeah, their website must be getting hammered. I keep getting "Something went wrong! [ try again ]"
OMG, this stuff is so crazy. I mean i am making music by myself, it is mostly for fun. But this Ai. I mean a whole son in about 1 hour. I need at least 2-3 days, and thats only if i have an idea. WOW. The Future is AI, for sure.
I'm impressed by how AI evolution keeps getting better, improving tools. While udio has its flaws, being in beta, I believe it will evolve even further over time. I even tried creating my own music, and I can say it has been an interesting experience.
I'm excited to try Udio and it does seem to have some innovations on Suno (I love the ability to add an extension before and after the current generation). Personally, I wouldn't listen to any of these tracks and there are Suno tracks I would listen to so I'm not sure it's overall really blowing me away but we need more time to see what it's capable of.
from the terms:
"You acknowledge and agree that we may use your INPUT Content and Output to train or otherwise improve our Services, including our artificial intelligence and machine learnings model(s) related to the Services."
"The Services may allow you to download copies of another user’s Output, and similarly other users may download copies of your Output."
So the songs are not private here...
They are not. I'll circle back to them in a bit, but from talking to them-- apparently in terms of the licence, you (the generator) have it. They keep a clause in the licensing to be able to use the output for ad purposes, but that's it.
Copyright is up in the air-- but that's an overall AI issue. If you generate an instrumental track and then sing over it, obviously that's yours and won't appear in any Ad material.
Again, at some point when things calm down (it is launch day) I'll talk to them again and get answers for all this stuff. In the meantime, just have fun generating some songs! (If you can get it to work today!)
@@TheoreticallyMedia Yea I just went through the terms and they do mention a private mode. But I couldn't find any privacy function on the interface. So as of now, sadly, with no private mode available I can't touch it.
Wow, I've discovered covers of existing songs and melodies there, probably from one of the developers, because I haven't been able to discover this feature anywhere yet. So this is already technically possible! I hope this will soon be available as a feature, but then of course it will quickly become difficult from a copyright point of view. So we will have to hope for the rapid appearance of an open source alternative!
Delving into storytelling and video exploration, VideoGPT subtly refines my content, adding depth and professionalism to my creative projects.
And removing all personality
Surprisingly good, "prompt" *70s aggressive electric guitar southern rock*
AI did it again, making a believer out of me 🔥Udio Rocks
7:06
That's pretty awesome!!
It's funny, I kind of slag on it in the video, but it is also the one song from the video that keeps getting stuck in my head...so...there is something to it!
So I've been experimenting with Udio for the past day and.........
I'm quiet impressed! 👍 👍 👍
The only glitch is when I got a warning about my custom lyrics being forbidden ( I had ChatGPT write me lyrics for a song about the troubling politics in the US in 2024).
Great video, bro. Can't wait to try out Udio. Love your taste in tunes, btw. Keep rockin'!
This is awesome! I’ve written lyrics for years and can feel the songs but but can’t sing or play well enough. Now I can create!!
yeah and only you will hear it. because every person can make this and maybe the ai will write better lyrics than you
@@Ruedi-ol1ul😂 i don‘t get how people don‘t see this. It is not a Tool it replaces.
@@flyversusfly76what’s the problem, I’m old enough to remember the start of MTV, video killed the radio star, Also people kicked off about the internet, MP3s even TH-cam, personally I think this is brilliant and can only get better, I’m a session guitarist, I can now make the music and video I want and do it all at home, just look at the music scene over the last ten or more years, A I is definitely a good thing.
@@hardlines2635 it‘s cool that you see the potential. Me personally I think why should anyone pay for a session guitarist when I can Click some buttons to instantly have 10 great guitar riffs „in the style of“ fitted right my Song? I think there will be a niche for human music as an Art form still.
It’s depressing that AI is allowing people to cut corners like this instead of actually learning the skills required, all while completely devaluing real musicians who have spent years mastering their craft
It occurs to me that creative enterprises are becoming less like noodling around for inspiration, and more like shopping where you are looking at a bunch of stuff and selecting what resonates with you.
I don't know if it's your fault for sending so many people over, but it just keeps saying there is an error and to try again later. "Generation is under maintenance" . I will give another try later. They probably need to update their system to handle the traffic.
This is my favorite AI channel these days. Thanks for keeping us in the loop!
This feels like it could be a Suno killer. My first generation was amazing! Thanks for reviewing this. I hadn't heard about it until this video.
The Frank Zappa part was great, I'm a big Steve Vai fan so that part made me smile. Good stuff.
Have you seen the Steve Vai documentary on TH-cam?
@@AlanBerry yep so good.
@@RetropunkAI thanks man. I made it. 😀
That's wild and... the end.of platforms who offer premium "royalty free" mediocre music for adverts etc... for non musicians. But that's life. The Zappa part was just crazy...
Oh yeah-- If a company like Splice isn't already working on something like this, then it'll pretty much spell their doom. Come to think of it, I should probably cancel my Splice Subscription...
I don't understand I got into the beta for 5 seconds now when I try to get on it says error
So far Udio seems interesting. Still a long way to go though but for sure a step in the right direction. I like it so far. Competition breeds excellence. :D
Are the videos you play with the song part of Udio?
They’re not. I generated those in Runway Gen-2, just to have something to look at, as opposed to just seeing the Udio screen.
OK, now that’s more like it! It just might be better than Suno! Guitars sound better than Suno and I’ll be putting it through the ringer over the next few days. Thanks for the quick update from yesterday. Once a paid plan is ready, pricing won’t matter as far as I’m concerned bcuz this is amazing stuff, exciting times!
As a long time guitarist-- the guitar sounds are INSANE. Not just the shred stuff either, I killed the "Bon Jovi Blues" song, but it had a pretty rippin' solo toward the end. Like, I was honestly thinking: I want to learn that solo!
@@TheoreticallyMedia - Same here. That was enough to convince me to sign up immediately. And after hearing about stems and comparing the GUI with its competitors’… I died and went to heaven.
After playing with it a little bit, seems like Udio has better overall sound quality, but Suno has much better composition.
They'll each get better at both in time with user reinforced training though.
From what I've heard "in there" (I can't test it at the moment, the website is under maintenance), IN MY OPINION, sonically, in terms of clarity and complexity, it is clearly SUPERIOR to Suno.
I managed to test it. Jesus Christ. When this "thing" is sufficiently trained and perfected, the phrase "composing music" will have another meaning.
Thanks Tim! Great video and nice to see that your music taste is great (Dream Theater, Zappa etc. Wow! Just my taste as well). Really enjoy your humorous way of thinking and making videos. btw we hae some commonality in comics as well as I´ve done them too (found on my channel as a youtube video). Thank You!
You really know a hell lot about music. I am very impressed with Udio. It's def not there yet but it's approaching at what seems like light speed. Very curious where they will be say half a year from now.
The tech is really cool but as a musician that doesn’t play shows…this is disheartening.
Why would any musicians support this? This will literally replace us and put us out of business.
And this is only version 1...
totally understand-- but in my mind: Because no one will ever pay to go see a prompt play live.
This is going to disrupt Stock, Type-Beat, and the Loop producer arenas...but, other than that, I just see this as a win for musicians.
If you have an idea, you can sketch it out with something like this. Or just generate drums only for a drum loop-- or (eventually) feed is a half finished song to see what it makes. It might be good, it might suck-- but, it'll get your head rolling.
It isn't going to replace us. No machine can survive the amount of beer that gets spilled on me during a gig.
Comments that are critical of Udio, such as @CypiXmusic and @javier_afonso (the latter of which was really eloquent and to which the channel creator offered no response), are actually being removed. Shocking, right? Anyway... Artists, from here on out, let’s go on a modified data strike that should make everyone, and I do mean everyone, happy. CREATE NOTHING NEW. Until we are adequately protected, let’s take remix culture to its most extreme ends. The pro-AI “art” community has told us repeatedly that there is no such thing as originality, everything is derivative, etc. They clearly believe we’ve reached the limits of human creativity and somehow feel qualified to tell us that art has become “stagnant.” (Never mind that these art whisperers include the same tech moguls who have fed the masses mediocre art through algorithms for decades.) Let’s not fight this strange alliance between billionaire tech moguls who are somehow suddenly democratizers and everyday people who believe themselves to be decent (or even progressive) but have no qualms about stealing from artists, including the majority of whom are poor or middle class. Let’s instead carry out their self-serving logic to its ridiculous end. Let’s turn the other cheek by doing what they’ve told us to do so many times: Be closer to their definition of “creative” and “adapt” as they have. Until artists are adequately protected and AI companies and grifters cannot scrape copyrighted works without artists’ consent and without crediting and fairly compensating each artist, stop feeding the machine. Use your advantage, which is skill. Make friends with and collaborate with other human artists who are against us being exploited to train generative AI. (This may be the time when you find out who your real friends are. Collective suffering brings oppressed people together. Unite.) Create remixes of each others’ works. Remix the remixes, too. And so on. These will be protected by copyright while AI outputs are not. Your skill will enable you to do this somewhat efficiently, and you will be less attached to these creations than you would be to the work you “dared” to call original. That said, never stop working on art that you truly love even if you don’t release it or only share it in special circumstances where the cost-benefit analysis makes sense. If you like creating original art, this kind of data strike means you may have to do some work you do not love (derivative work) but it will still be art (you will still probably like and enjoy it) and not a day job you hate. If you want to use AI (as you are constantly told you should) to ensure that you stay on top of the tech or to reduce the amount of original, human work available for corporations to scrape (and thereby contribute to the logical consequences of AI feeding itself its own ****), there are ethical models. (See Fairly Trained.) Also, keep pressuring your elected representatives to end to this madness.
UPDATE: Udio has already fixed the issue with consistent choruses that I mentioned below. It can now understand what's in your track within a 2 minute window whereas before is was only a 30 second window. That's impressively quick for support. Now, for me, this is now beating Suno, especially since Suno has run into issues with version 3. I've been able to easily extend Udio clips to over 3 minutes. At this point, I'm just waiting for the musician tools and an increase in clip length. Stems, or hopefully every track separate, will be prefect and that seems to be on Udio's radar.
***
Having used this for a few days, there are definitely some things missing I'm sure others who write songs with lyrics may have noticed. I wasn't able to get consistency. Not even close. The extend feature often results in chaos. trying to get consistent verses and choruses isn't easily achieved. What I would prefer would be to see an app where you get to work on single sections without having to worry about fitting things into a 33 second window or some other number. Generating things and crossing your fingers isn't working for me.
For example, it would be perfect if we had an option to work on the intro, verses, pre-choruses, choruses, bridge and any desired instrumental sections separately. Once and initial selection for a section is made, they should remain consistent with a scalable amount of variation in the music and melody via perhaps a slider that determines the amount of variation desired. Once worked out, we should be able to chain those sections as desired within an "arranger" that provides smooth transitions between the sections. And if we don't like that arrangement, we should be able to change it by re-ordering the sections and have the app automatically fix the transitions between sections to match. That would be much closer to how many songwriters work things out.
Maybe this is asking for too much in the near future, but if this could be considered, I think everyone would be happy. For now, I'm doing a lot of offline editing and a lot of generating that wastes credits.
This looks very similar to Suno AI. I’ve spent two nights creating new songs from lyrics I’ve had waiting for years to put to music.
site is falling apart cant get it to load
probably getting slammed.
I'm sure that server is getting SLAGGED right now. haha...the hug of death!
It's working awesome for me so far.
That's great to hear!
Suno is universal, the voices are much more clean. I will play with Suno for a while and let it grow. Thanks man!
After reading the positive comments about this, I thought I was the only one that thought Suno sounded better. Yeah something about the voices on Suno being more crisp
You obviously have no idea what you're talking about. I can burn 2k tokens in suno before I get a good voice that doesn't sound robotic.
Udio sounds way better to my ears, a lot cleaner than Suno... actually I thought the difference was drastic on first impression but I'll bring them into ableton and a/b them.
can you use a made and downloaded song in a youtube video? or is that not allowed with udio beta? or burning your mp3 you made with udio and burn it to a cd and share it with friends?
You should be able to. As it currently stands, there is no copyright restrictions on AI Generated music.
It's diabolical how advanced AI in 2024 has become.
Boggles to think what 2030 looks like.
@@TheoreticallyMedia You are not going to like it. Or maybe you will
The AI's Jacques Cousteau reference at 9:53 is probably from Wu Tang Clan - Da Mystery of Chessboxin. There is a lyric with a similar reference by Ol Dirty Bastard in that record.
So what I get from UDIO is that it's to Jukebox, what ElevenLabs is to Uberduck; Like it'll still make music with better audible fidelity and consistency than Jukebox, but it can't generate output from pre-existing materials/attempt to copy one artist's specific styles (like that one Bruno Mars sample from the Jukebox library)
I "created" some great songs with Suno, my only regret as a musician is not having anyone to play them with :-(
@TheoreticallyMedia Good stuff as always, Tim. If you had asked me 3-4 months ago how long it would take to get to this level, I would have guessed at least 5 years. It's amazing to think that THIS is where we are today, and it will only improve. Of course, how we, as musicians, ultimately utilize the tools will be the real challenge.
It has no support for other languages, only English, which makes Suno infinitely superior despite the more creative results of Udio.
That's a VERY good point! Udio is just in 1.0 Beta, so give them some time. I'm not a "this vs this" kind of guy, so I'm just happy we've got two really good music generators!
By the way the best (for me) methode for making AI music is asking Chatgpt for MIDI files. Of course Is a "crafty" way but the results are wonderfull
Where does the video in the bottom right hand corner playing along with the songs come from?
Oh, that's just some stuff I put in so it wasn't just a static screen. Random videos, Animals As Leaders, and a band called Turnstile for the punk rock section.
@@TheoreticallyMedia Oh and here I was thinking I had somehow missed a cool bundled music to video feature lol!
All the people who worked so hard to make music over so many years, now some tech bros with no talent can profiteer off of scraping and copy-pasting their work in an icky digital mish-mash. Yippeee! Definitely their mothers can be proud, as they finally solved the problem of the average musician just being too rich, and the average tech bro not rich enough.
Artists, from here on out, let’s go on a modified data strike that should make everyone, and I do mean everyone, happy. CREATE NOTHING NEW. Until we are adequately protected, let’s take remix culture to its most extreme ends. The pro-AI “art” community has told us repeatedly that there is no such thing as originality, everything is derivative, etc. They clearly believe we’ve reached the limits of human creativity and somehow feel qualified to tell us that art has become “stagnant.” (Never mind that these art whisperers include the same tech moguls who have fed the masses mediocre art through algorithms for decades.) Let’s not fight this strange alliance between billionaire tech moguls who are somehow suddenly democratizers and everyday people who believe themselves to be decent (or even progressive) but have no qualms about stealing from artists, including the majority of whom are poor or middle class. Let’s instead carry out their self-serving logic to its ridiculous end. Let’s turn the other cheek by doing what they’ve told us to do so many times: Be closer to their definition of “creative” and “adapt” as they have. Until artists are adequately protected and AI companies and grifters cannot scrape copyrighted works without artists’ consent and without crediting and fairly compensating each artist, stop feeding the machine. Use your advantage, which is skill. Make friends with and collaborate with other human artists who are against us being exploited to train generative AI. (This may be the time when you find out who your real friends are. Collective suffering brings oppressed people together. Unite.) Create remixes of each others’ works. Remix the remixes, too. And so on. These will be protected by copyright while AI outputs are not. Your skill will enable you to do this somewhat efficiently, and you will be less attached to these creations than you would be to the work you “dared” to call original. That said, never stop working on art that you truly love even if you don’t release it or only share it in special circumstances where the cost-benefit analysis makes sense. If you like creating original art, this kind of data strike means you may have to do some work you do not love (derivative work) but it will still be art (you will still probably like and enjoy it) and not a day job you hate. If you want to use AI (as you are constantly told you should) to ensure that you stay on top of the tech or to reduce the amount of original, human work available for corporations to scrape (and thereby contribute to the logical consequences of AI feeding itself its own ****), there are ethical models. (See Fairly Trained.) Also, keep pressuring your elected representatives to end to this madness.
i tested this, I'm just amazed, the flexibility, being part of Theoretically Media 3X your capability with AI. Such a great commentator and brings great Value with every video.
I thought Suno Ai was rad... I mean, V3 is still great but ound? Holy molly maceroni!!! That's another level! And the control is also more advanced than with Suno! ... Crazy.
I'm really speechless... I think here too, as with every AI, you're bound to come across a repeating pattern at some point, but still! Just the fact that mam can combine completely absurd style directions with each other relatively easily and without cramping AND that it still sounds good or at least interesting... Let's see if I can still get my metal song realized the way I want it to be :)
Metropolis Part 3: publish it. Would love to hear the rest of that one. I am wondering where to go from here? I am thinking: gen tracks. Slice and dice into segments/stems and then onto DAW for composing. Thoughts?
So, Metro is actually on the "Trending" page-- if you can get in (ha, I know the servers are slagged right now), you should be able to search for it. It is GOOD. I only played that clip, but I really had a hard time deciding where to clip it!
Also: Stems are coming from Udio! Not at launch, but it is in their roadmap!
You sir are always on the bleeding edge. Thank you!
My Udio generations are taking a long time... what were your average generation times? (watched your whole video, still waiting for my 1st gens to complete, haha)
Update: 1st 2 generations took 20 minutes.
Next 2 took 33 seconds! 🤯
Sorry to bug you again, but you by far explain things better than others so you are stuck with me :) Two questions (1) I don' t get the work flow....When ready to create the song, I click the CREATE BUTTON --- when udio is done creating the song, next I clicked the EXTEND BUTTON. I'm satisfied with the song, so I'm done and ready to do a new song. BUT HOW DO I GET BACK TO THE CREATE BUTTON? I seem to be stuck on the EXTEND BUTTON and could not figure out how to get the CREATE BUTTON back. Instead, I exit Udio and open it again to get the CREATE BUTTON. (2) second question. Is there a way to pick the singer you like. I love the female singer I got on my first song and want to use her in other songs but keep getting singers I don't like. Thanks for your help. I love your videos but as a retired vet on a fixed income I'm always counting pennies.
Woot I'm in! It took a day to get into the beta and 30minutes to generate a song but damn the quality is great!
1:54
Definitely crank it up.......fan of Prog Rock here for decades 👍 👍 👍
The "Zappa" section has an interesting bit. I call it Zappa, but if you know your prog, I also hear Yes in the vocals! I didn't mention it in the video, or I'd end up spending 50 minutes rambling about prog, and then someone in a cape would show up to do a keyboard solo. hah, no one wants that!
@TheoreticallyMedia
😆 🤣 😂
Yeah, that Zappa-ish song is interesting indeed. It shows that Udio can produce listenable
off-tempo weird compositions too.
That seems even more challenging...... but I'm not a musician so I could be very wrong on this.
It made me smile when you smiled at your Push me to the limits part. Just a stupid "I'm enjoying this!" smile...lol.
sure, i can have an AI to compose songs for me, another one to mix, and one to master.
then i could spend my days on the beach with all that free time! lol
in the future you will probably generate one single song in your prefered style, and then have an AI agent create entire albums and a virtual band based on it with music videos included
Perhaps theyll sample your voice and create bands that you can own the rights to basically based upon your description of structure and aounds, hord progressions, instruments, etc then your band has a sound and style created with an albums worth of songs ready to practice and go on tour with.
Maybe even sample your guitar and if youre sloppy or slow they could fix it up or make something similar as mentioned before... etc
I think that'll be something we'll see in a Pro version down the line...
How did you generate videos for theese songs? I see option "download video" but does not work :(
I have a question? I am a paid subscriber with suno. With that all content I create is owned by me and can be sold commercially. Currently Udio is free. Does that mean whatever I create they own the rights too? If so I would rather wait til they have a subscription plan similar to suno.
their servers are getting hammered this morning
They still aren’t working
So, should I stop my chord studies? I mean, what is the point? A life without effort is pointless. You know someone's going to try and sell these songs and then the lawyers are going to drop on this like a ton of bricks.
I actually am shocked and stunned this time. I'm working on extending a Nirvana / Linkin Park / Metallica song
description alone makes me want to start a mosh pit!
The only thing that could make it perfect ,is to have "stems" automatically. And the sound quality will be "wav"
That is something they’re supposedly working on for a “pro” version
Yeah, like you mentioned at the end. If they really want to help artists to make something cool, there need to be features like: prompting the BPM of the song, the time signature, the key of the song, maybe even the chord sequence and of course stems and maximum control over adjusting certain parts (maybe even per bar). A system of adjusting the details is necessary to co-create with A.I. Sound wise it seems a bit better than Suno V3, but the editing features are too limited. The devil is in the details... I want control. That's where you can raise the quality of a track. I wonder how long until I can upload a rough demo, make it into a finished song and then when I'm happy prompt: "Mix it like Chris Lord-Alge."
Your comment made me laugh because I have one of the Chris Lord Alge plugins and no matter what I do-- it isn't coming out like that!!
Yeah, all of that is forthcoming. I'm following up w/ them on everything that's being brought up in the comments. I'll be back with another video at some point-- a deeper tutorial dive and (hopefully) some more answers from them!
@@TheoreticallyMedia Ha, that's great to hear Tim!
@@TheoreticallyMedia At this point not only spootify must be worried the Big DAW dogs (in the likes of Cubase) as well ,I can certainly imagine future versions not only take chords, melodies but you will be able to do sequencing, add virtual instruments and the whole production by just prompting ,full control is definitely every songwriter dream when it comes to AI generating music
After using the workflow for a few days... We were totally wrong about wanting one gen at a time.🎉
I'm trying to generate a country song ( I tried with a lengthy description and with a short simple description), but it keeps saying there was an 'error'. Not sure what to do. Did that happen to you?
Lols i'm an actual suno ai creator lol I write my own lyrics and songs and I have ai perform for me on my ch lol. Good video bro . New sub ❤
Oh, I love Suno too. And they’re about to drop a pretty major update. Suno and Udio are both arm wrestling for best music Ai, and at the end of the day: we all win!
Never thought I'd get a Jawbreaker reference in an AI video!
FIRST TO CALL IT!! I'VE LITERALLY BEEN WAITING ALL DAY FOR YOU!
haha, we are now best friends, of course!
@@TheoreticallyMedia hell yeah!
Anyone else getting malicious website warnings when going to Udio's home page?
I find this pretty interesting. As a musician and recording artist, this is largely irrelevant to my work- I am not going to stop doing my creative work simply because someone can press a button and make a song. Where it does become intriguing is as a license-free source of samples to chop up and work into my original compositions. I’ve always been bothered by the idea of taking other people’s music and claiming it for myself as a sample. Potential issues with training data aside, this feels like a more ethical way to produce unique and royalty-free samples.
Yup! You got it right there! I've been doing a lot of the same since like...well, pre-Suno. Back when it was still RiffDiffusion! And hypothetically, you could generate just drum loops with something like this. Or just piano chords.
One thing for sure: If Splice doesn't get it together, they won't be around much longer.
No offense, but I am sensing an overuse of superlatives in the video titles of this channel.
Nothing that is this easy has value. Never had never will have. It means music will loose its magic and value totally. If you cant see that you where born yesterday.
Hi, just tried it - really great sound quality. Would you post a feature request for me (can’t find a way to do it myself)/? I really wish I could post a melody (as midi or just written) and have it arrange a song in a certain style with this melody. As amateur musician this would make it instantly more useful and a learning tool at the same time.
Very cool, the sounds are great and I was literally talking to my wife about this topic last night. Currently the generation isn't working for me (something about being under maintenance) but listening to the music already there is nice.
what novie is that that you are showing in the minute 3:42?