Before you comment, let me clarify my viewpoints on this: 1. This technology is going to exist no matter what we do, we need to learn how it works and figure out how tools like this can help us as music creators. 2. Someone needs to create a system that properly compensates artists who are used in training data, and it needs to be opt-in only. Nobody should be used as training data without their permission. 3. Governments need to act fast to impose laws for how music is used in training data. Not everyone will follow these laws, but they should still exist. 4. This isn't going to replace artists. Don't panic. Fans want to connect with real artists doing cool things and people mass uploading AI music can't replace that (at least not soon). This will however likely screw over a lot of artists making more templated music such as lofi, meditation music, corporate background music etc.
Please read the terms of service or check out the Top Attorney episode on the TOS. You can’t copyright or have sole rights to the masters. That makes this tuff useless.
With other AI´s that create videos and photos you can create a whole AI person, we already see that with AI influencers. Now the only thing you need is a hologram of that person and you can do live shows, or virtual shows (apple vision etc.). I belive that we will see a lot of those "artists" in the near future (5-10 years). Think about the budget cuts for major labels, they don't have to share with an artist anymore. Hire a few designers and prompt writers + a producer to touch up the generated music and you are good to go.
Governements are no longer the answer because they are in bed with the big industries, and they have been for a while. What will happen is already decided, but it must seem to come from us. We need to choose it, even if it would not normally be our choice. For that to happen, we (the people) are going to suffer it first. It’s a long con false flag operation. Why do governments agree with this? Because they have been fed a certain apocalyptic narrative that shows what will happen if they don’t. (See Daniel Schmachtenberger’s explanation of this issue for more info.)
Partly agree. There are also many good producers who work in the background but don't connect with fans whose music is not "templated". Their economic existence is definitely under threat.
But there is a point about people, they are always something new, and AI is not able to innovate, to create something original, these sounds are generated from models created with songs by other artists, that is, generic music, I believe that, many people will lose space for sure but original music will be even more requested.
New small artists are the ones drawing the short straw. There's going to be countless AI songs out there to compete against, and it won't even matter if you use AI or not yourself. You're still going to get swamped against the vast amount of new tracks out there. The outcome for new artists will be more PR, more social media content, more non-music stuff to get yourself heard. Established artists will be affected too of course as there is only so much ear-time of listeners out there, but I'll guess they'll be affected a lot less. Thanks for your thoughts.
Yes, agree. think that as the market becomes oversaturated with AI, real things like live concert performances / live stream performances will become more valuable.
@@GhostWriter_Music Yup. Musicians will have to work much harder making better music, more content. My head is already spinning. I might need to see a priest about that.
@@GoranBackmanMusic no longer will a studio or artist be able to just throw money at a problem and it get them a bunch of fans and profit, because that will be expensive.
Ive been in a touring indie band from 2013-2016, and now a full time music producer working for trippie redd, kehlani, machine gun kelly and more, and I have to say Andrew nails it when he says this tech is both amazing but also terrifying. I don’t think we have any idea how this is going to change the landscape within 3 years let alone 10!!! I don’t think real music is going anywhere anytime soon, but I 100% agree it’s good to keep an eye on this technology and educate ourselves. Great video as always.
I just came across this video and found it to be very informative! Thank you! I have never written any lyrics or songs in my life before finding a RUclips video about SUNO. I have been using SUNO AI since September 6, 2024, and began writing lyrics using bible verses or things I have written in the past, and creating music videos. I truly believe SUNO AI is giving people the opportunity to be creative and it doesn't cost a lot of money to get their creative products out there. Thanks again for a very informative video!
That's a good point. Meta released an open source music generative AI tool a couple years ago, and they were very vocal about the fact they trained it on 100% properly licensed music. It's an awesome marketing angle to be able to say not only is your AI tool amazing, but it's the most ethical one on the market.
Been telling people this the whole time. People seem to think ai has no soul so it won't take off.. they are clueless and don't see the writing on the wall. People will generate the songs and then add the soul to it when they modify it. The market is already over saturated and it's going to get worse with ai. Big artists will use it and now they don't have to share publishing with the people who were writing their songs cause they can now create their own songs... it's going to be crazy.
I am strictly a lyricist. Not the best, but I write a pretty decent song. Can’t play, can’t sing. Been writing for about 15 years. Prior to AI, I would finalize my lyrics, sing my song into my voice recorder, send it to some Nashville studio, pay them anywhere from $300-$1000 and wait 7-14 days to wait on mymasterpiece. Was always happy with ny final result, but there was always something you wish was different. And over the last 15 years pitching my songs, I’ve had to endure ego after ego, gatekeeper after gatekeeper. Disputes, you name it. 3 months ago I stumbled on Suno and paid for the pro service. Inputted my lyrics, made my prompts and let’s just say what they generated were literally 100 times than I even envisioned for the song. So now I pay a small monthly fee, and get literal masterpieces in my living room in seconds after completing my song. No egos. No drama, No paying $750 to wait two weeks from some premadonna who ghosts you or tells you what your song needs not the other way around. After a few days of what I stumbled upon and seeing the results I was so inspired I proceeded to write probably 100 songs in a short amount of time. Songs about anything and everything, with my own lyrics. Beautiful stuff. Wrote songs for every person who’s ever been in my life, and moved them and their family to tears. Now have a significant music industry vet in my corner, getting interest in my songs that contain my own lyrics. The lives that these songs I’ve created with Suno’s help have literally moved so many people, not just myself. But everyone wants to tell me ai is not good and the music isn’t good. I’ve had songs placed before when I worked with humans, had my share of compliments.Songs picked up by the NFL, NHL, etc. And my songs were pretty damn good back then. Now they are mind blowing and now for the first time I have a fighting chance to compete with the big boys. Right now I have a 30 year Ivy League entertainment attorney submitting my songs to A-list artists. I use my songs as work tapes. Currently have songs out 20 major artists. All 20 might pass and I might never get a cut an again. But I tell you I have about 100 of my songs on my playlist and they are the sound track to my life. With that said, people who say AI is soulless and the songs suck and, just aren’t paying attention or some musical dinosaur felling like they are about to be extinct. I ❤ ai music. Who the heck doesn’t want a music studio in their home that doesn’t ghost you, lock you out, send you duds and keep your money. I get what I want and I get it when I want it!
I made AI album a year ago. Been saying for a while producers and artists gotta curate a sound and build and market that. I bet Spotify will eventually generate AI music that matches YOUR algorithmic tastes. And the royalties will go to them 100% I'll add on top of this. Soon artists won't need to pay producers when they can generate a million type beats until one is close enough. They will own 100% of it. Producers. Adapt now. Invest in artists not beat sales.
100% agree there could be a day Spotify adds an AI music feature that generates infinite AI songs that adjust based on what you listen to most, and what you skip the most. The good songs get turned into permanent tracks that they could then recommend on other playlists. Honestly this is probably the most dystopian situation I can imagine, so I really hope this doesn't happen. I feel it all relies on if the average consumer enjoys the tracks or if they reject the idea of it. A lot of artists would push back against Spotify doing this too. But from Spotify's perspective, artists and labels take 70% of their revenue. If they can cut any part of that out it significantly increases their bottom line. I feel like this particular aspect could be illegal though since they're a platform - just like how Apple isn't allowed to promote Apple Music in their store more than Spotify, despite the fact they own the device and the store. Also just like how Microsoft got in trouble for pushing their own browser. Congress could view Spotify pushing their own internal music as stifling competition from the rest of the industry since they have so much marketshare.
Something about blasting artificially generated material in on the same level with human created just seems morally wrong to me, music or otherwise. That’s the larger issue with all this I think. Will they allow robots to compete against humans in sports ? I don’t think the training or curating or “knob twisting” is particularly relevant. Michael Jordan playing against a kindergarten basketball team doesn’t add up
I can see this happening. It's very likely we are headed towards a scenario where everyone's life is AI curated in every aspect, as far as entertainment. One way to slow this down? The AI companies need to be sued, and they need to stop being able to scrape the internet for data. The tech is only relevant so long as it is able to scrape all the current data.
@@AndrewSouthworth Spotify already commissions music from producers, I've met some of these guys personally. They get paid upfront fees (e.g. $1k) to churn out EDM for Spotify's editorial playlists and waive all rights and ownership of the music. These songs have millions of streams and fake artist profiles. AI music is the logical next step for Spotify et al and I fear the average listener won't care where it came from, they'll just listen to a playlist and go about their day. Human made music will be niche - like if you were to go buy organic produce at the farmer's market.
I think you're absolutely right, this has probably existed for at least a decade and the public is just finding out about it. If we have access to this for a $30 dollar subscription, just imagine what they have access to in a million dollar recording studio.
I make AI music, as a hobby (i write my lyrics) i just choose a style and also i edit what i dont like, Music is beautiful no matter how its made, the future is here, the same was said about digital keyboard, with loopers, this is just much more advanced, Nothing takes away other peoples music ever! true artist will always be better, even as an AI Music creator i cant compete with real music
00:00:00 - Introduction to AI-Generated Music and Its Perception 00:00:28 - Creating a Song with AI: Demonstration on Two Platforms 00:01:15 - Song Comparison and Capabilities of AI Music 00:02:37 - Exploring AI-Generated Metal Music 00:03:10 - Understanding the Quality and Limitations of AI Music 00:03:40 - The Ethical Considerations of AI-Generated Music 00:04:41 - Training Data and Copyright Concerns 00:05:16 - The Future of AI Music and Its Impact on Artists 00:06:08 - Potential Solutions for Fair Compensation in AI Music 00:07:02 - Generative Music and Ownership Questions 00:07:36 - The Role of Artists in AI Music Creation 00:08:02 - AI as a Tool in the Creative Workflow 00:08:32 - Conclusion: The Future of AI in the Music Industry 00:08:59 - Final Thoughts and Viewer Engagement
Parody music generated by AI has taken off like crazy, there is an AI artist on spotify with 200k+ monthly listeners with music from suno, and youtube plays in the multiple millions.
Why should this be monetized? Little to no effort, little to no creativity, hardly any time to generate. Reddit is literally filled with discussions on people very upset that AI content creators on RUclips are trying to grift or get something for nothing.
Oh my lord. It really is insane what it can do. I think the creators in the music industry most affected are going to be the producers or anyone who programs music, be that EDM etc. The winners will be the actual musicians who can play instruments live and are willing to go on the road to do so. Even if you use AI to write the songs for your band, it cant get up and play them for you. You will still have to that. And DJ's? The Ghost producer days are numbered for sure, but DJ's are curators at the end of the day, so regardless of whether they make the music or AI does, they still decide what to put their name to to and play at sets, so don't see a big change there.
Yeah it's kind of like that newish South Park episode, where all the white collar workers don't know how to do anything anymore and so the handy man workers become billionaires because they're willing to do stuff with their hands and learn how to fix stuff haha
I could be wrong, but think with me, everyone is going to have access to AI Music, and it's not able to create 100% original music, so everyone is going to have the same type of music and they're all going to sound generic, and today the artists that emerge is because they have something unique, different, they deliver something innovative, and the AI is not capable of that, so about Ghost, it is possible that for some artists they stood out and a lot of them don't know how to create music, they can turn to Ghost who know how to create something innovative and that makes sense in the project of these artists does it make sense or am I being too optimistic? Because it's impressive what AI Music does, but I've only heard generic sounds so far, I believe they will improve the technical quality of the sound mix, but "creatively" think they don't go beyond that at least for a long time
@@Gutz-po9xf I can promise you that you are not correct. The top tier AI output is absolutely as original and inventive as the very creme de la creme of inventive and original music from the past. You just need to look a little deeper.
@@AndrewSouthworth great episode and that is exactly how i see the future playing out. BUT upon a foundation of universal basic income; where people can then earn extra credits based off effort / exchanging / deals
Winners are vocalists and lyricists. The copyright office better wake up and stop claiming everything made by AI is free to copy because people will lie about how something was made, and lawsuits will fly if people rip off other's lyrics and also voices and vocal characters. Losers will be generic composers who have already been losing for over a decade anyway. The film business is also imploding. Eventually govt will have to step in, but in the USA, we've got a rising tide of fascist religionists who hate govt support for the arts anyway, so best idea is to move to another country if you want to be a creative.
On average, popular music has already become more polished, less natural, and even somewhat ‘robotic’ in the past 20 years, and people seem to have accepted the often ‘plastic’ feel of it. That alone should be enough to indication that A.I. will definitely replace a lot of musicians. But only those who don’t understand how A.I. works, and also don’t understand humans, will say that it will never be able to replace musicians. The key here is not A.I. though, but humans. Although we have made scientific and technological progress, as humans we have not truly evolved. We are using the same level of thinking and consciousness that we have been using for at least a couple of centuries now. It is precisely that kind of thinking which has led to these technical advancements, but which has also kept us away from real human progress. Once we start to realize this, society will start to change rapidly and it will choose a different direction. Hopefully, this realization will not come too late.
That 40s thing is just an effing joke ... the lyrics don't make any sense and it's just so generic. Gimme a break. It's crap. But so is MacDonalds and plenty people eat that - but they didn't put good restaurants out of business ...
Most casual listeners of music, don't care about how the music is made. No music producers or singers make songs without being inspired by someone else.
Just think how much AI has lifted the bar on vocal skills, musical skills, arrangement and production . The worst you get is some cheesy mess, the best is truly scarey good.
give it a little more time! This is really a problem for all artists! I don't think there's much we can do. The music industry was already crappy after Napster..the death of cds, then Spotify royalties and play revenues..now this? it's just getting too much to bear. I wonder why should we keep trying? it's really serious. I bet they're already musicians pumping out lyrics or albums really that they don't have the skills to, and then cover those songs live, I mean who can tell? I struggle with a blank paper to write lyrics, come up with the music and putting stuff together..now you can only be a cover artist ..hell..you don't even have to be a musician at all. Im not against technology, but this crosses the line. And I don't believe there's a law that can control this from happening. They could backup automatically every AI generated music and use the existing means for copyright control scanning..but how about I record the song generated all over on reaper and then upload to CDbaby? this os "cheat gpt" and no way to know. A band can release 6 albums a year or 2 each month while the honest guy will come up with one, maybe two in a year if they're really inspired. Not fair and quite horrifying for us all.
I honestly think it will be the same as with AI art. People will use it, creators might use it but ultimately it will give more value to human made art. People need that human connection.
I agree. If everyone can make decent art with AI, the people that make amazing art from scratch feel a bit extra special because they didn't NEED to do it from scratch.
I'm trying to see if a large portion of music listeners and creators are going to avoid playing AI created songs, to the extent that they petition spotify or soundcloud to filter out A.i created content so they can only listen to real artists. This could put a big cut on AI stream count and numbers/royalties, but just thinking that most of us never asked for ai music. I'm open to thoughts on that, as I don't know if that would work or how it would play out, I just know I don't want to listen to ai created content... maybe just an internal bias I'm having I can't explain.
@@AndrewSouthworth They actually did need to create it, because AI can not create what a human would create. A friend of mine proved that no combination of prompts could actually match his mind's eye. Creating art is something only sentient beings can do, and AI is not sentient. If I smash my printer, that can be considered art. If my printer falls off the table and smashes because it is shaking too much, that can not be considered art. The difference is in the intention behind the act. And no, entering prompts is not a creative act, as it lacks predictable results. Art must have CREATOR INTENTION behind every detail. This is why we do not consider a pitch to be a movie.
I don't think that simply entering a prompt and getting the output is art either. However as these tools evolve I suspect there will be plugins and tools created that assist artists in creating. For example, programming a simple drum pattern and then using a generative fill feature to humanize and add much more detail to the drums. When you think about generative Ai tools in this way it's a lot more artistic. There are also tools being added already that allow users to customize their AI creations using 'inpainting' for audio. This is still not very artistic but it's one step closer to giving the user more control of the final output.
Embrace A.I. It is only going to be bigger. Artists? What artists? They are manufactured now. Get a model, write some lines, and auto tune voice. A.I. does it better. Compensate? That's like Crayola suing you, because you made a great work of art with their crayons. Connect with artists? Why? I just want good music. What about music that doesn't have any vocals? Smooth Jazz. Does it matter who or what made it? Btw, the music industry is trying to own the rights, and create their own A.I. That way THEY will be able to make music in the future and not pay a single human. They will own it all.
I downloaded Synthesizer V to help me complete songs that I had stored in a hard drive for years, but once I figure out the vocal harmonies, I intend on getting live singers to record them. So in my case, I'm using AI as a writing tool. To clarify though, the program doesn't create anything for me. All it can do is sing the words I enter using the melodies I create for it. My voice isn't great, so it's very discouraging when I try to sing them. Using the AI voice can also help me guide a potential singer on the vibe I want for a certain passage.
We're doomed. It was already bad, but now it's the fatal blow since we, musicians/composers, have become totally useless and the ultimate sinkhole for the average listener musical tastes. People endorsing this and making profit are the ones who should get deleted.
True it will flood the market and devalue all music. Also content creators are already using AI music/sound effects. Won’t be long until game studios, commercials, movies, shows and animations begin using AI music too
All of this doesn't exist without human performers and to your point in the description laws need to be passed on this thievery - "Gramophone days" is obviously modeled on Frank Sinatra, not sure how you missed that - I'm sure his estate would be interested in contacting the AI company that "stole" his voice.
That's a good observation. All AI generated music is basically "photoshopping" fragments of known songs and pasting them together. It's cheap and grotesque.
@@heavenseek And the funny thing is that AI companies call using our stuff against us "fair use". This AI fraud will be remembered as the greatest scam in the history of computer technology.
@@federicoaschieri That's not how it works at all. The AI is trained on what "Rock" sounds like and what "Gospel" sounds like. Then when you ask for a "Gospel Rock" song, it starts with pink noise, and starts EQing the noise until it approximates what a combination of "Rock" and "Gospel" might sound like. The more specific the prompt, the more likely you will generate something that sounds like something else. This isn't meant to defend AI music, just to combats the idea that it is copy and pasting parts of other peoples music.
Anyone who thinks AI isn’t going to destroy music is in denial. I have used it and it creates better music than I can make myself. The singers are perfect, the bass is clean and hits hard. It’s already better than what humans can create.
It may be better than what some humans can do now, but right now it isn't better than most pro musicians. There will come a time when it is, but in this moment it's only better than newer musicians / artists. I don't think it's going to cause much harm to real music. Many listeners do not want to knowingly listen to AI generated music. Fans want to see live music and build a connection with an artist. AI artwork has been around for years now, and it's producing photorealistic images or artistic images in any style. Most professional visual artists are still professional visual artists. The AI artwork mostly only took away the bottom of the barrel work.
I don't think AI music will replace Human created music. But AI music does make it easier for people like me to create music without the hassle of looking for band mates to form a band. I'm good at vocals, but again I got no band mates to play with. But when it comes to lyric ideas and musical style to use with AI, I come up with the best songs that has came directly from my mind. And as long as the AI music comes up with original sound with the lyrics the creator came up with, only the creator should profit from it.
I just recently took my lyrics to my song and used Donna AI to create music to it. The result was a song so moving that anyone that listins to it loves it. Now the problem is that BMI at this time don't know how to handle it, they say their working on the problem, and that I'm not allowed to register my "partial AI created" song in my BMI Catalog. But Donna AI says that I have the right to use my Donna AI created song anywhere I want. I am going to use it as a demo to submit to artists for recording. I just don't know how that would work for a label unless the songs music evolves in the recording process to be different from the Donna AI version. But everyone tells me that the song is perfect as is as far as the music goes, so I'm not sure how that would effect the label and or artists interest in the song.
@@RobertHampton747 Yeah, I can see the roadblock in doing that. I have some friends who are musicians, but they don't play the style of music I used to create my songs. I'm thinking of going to use CDbaby or Distrokid to distribute my music albums with.
@@RobertHampton747 true. I myself want to distribute both my albums digitally. The thing is my music might be politically controversial. So I don't know of distributers will accept it.
I love the analytical approach you have to things! Instead of "OMG AI will take away everything" you explain logically how we actually use generative tools already all the time. I personally think it's good for people who have no idea about how to make music and don't have the capacity to learn all that stuff (It's A LOT to learn) But for me personally making music is putting my soul into it, and I cannot do that with AI, because I love singing and putting together my own puzzle pieces. Maybe I consider using AI for parts of the music, but making a whole song with it... Idk maybe it sounds good but it will never feel like "mine"
Thanks! I feel like there are a lot of sensational opinions on both sides, but I feel like reality is usually somewhere in the middle. I can imagine a suite of generative AI tools that allow us to generate drum grooves that match guitar riffs, bass patterns to fit under chords, or allow us to generate 10 variations of a melody idea so we can more quickly find the best possible version of an idea. Or even just to kill writers block and give ourselves a starting point for a song idea. Or more sophisticated mixing and mastering tools. Not everyone will use them, because some people like doing everything from scratch. But it will enable a massive amount of artists to make higher quality music than ever from their home, without needing to hire session musicians, buy beats or pay for studio time. They won't replace talent, but talented people will use these tools to become even more proficient and efficient.
@@AndrewSouthworth yeah exactly! I am really tired of having to scroll through dozens of HiHat loops etc, instead i would love to have an AI that can generate variations and choose the best, similar to what MidJourney does. I generate an image, it gives me 4 variations and I say which one I want or I want more like that. Good examples that you said, so often I have a nice chord progression but I have 0 idea what kind of melody to play! Or how to make a nice Bassline to it. Hope there will be something like that for Ableton soon, I would love that!
What is training data for a song writer to sit down and write a song. I will tell you. Every song they have ever listened to in their lives. Isn’t that unethical to go back on life’s experience to create a song, especially if they do not go back and compensate all of the song writers who had a copyright on the music. So tell me what is unethical.
The future will be people selling their NIL for others to create music with their voice. Its coming for the movie industry also, it's just a few years behind. The future will have people making their own Marvel, StarWars, StarTrek or whatever... Streaming platforms will have a "create" center where people can make their own Avengers movies using the NIL of whatever actors sign the agreement. If you didn't like the new Star Wars movies, well guess what you can make your own in the future.
Right now, it seems to me that yes, small artists will discover that they are drowning in a sea of sound-alike AI generated content. No matter what ingenious novel style they try to come up with. So, that's going to be a tragedy for the development of genuine musical talent and creativity. But, also, it already seems to me that quality AI productions will be drowned out in a sea of middle of the road AI crap. Of course, that was always the big problem with the music industry. It turns out that it really is down to the terrible taste of most of the music downloading public. There will be AI generated gems hidden away on RUclips with only 3 views. And meanwhile I predict that the download charts will contain AI generated identikit Reggaeton and Afrobeat rip offs with the same rhythm, same basic bassline and horrible autotuned vocals. It's what the people crave most, apparently.
AI can make a good song, but AI can not make a great song AI by itself won't beat a human, but AI + human will beat a human After understanding this, i feel much better!
@@Recuper8 You are obviously not an artist and know little about music. A good song is good because HUMANS enjoy it. There are infinitely more chances of getting a hit song generated by an AI led by a tasteful musician/producer than one led by an average Joe who knows nothing about music.
@@Recuper8 AI is on track to hit a point where anyone can press the 'maek soong' button and make something 10x better than Mozart. But I think the most artistic creative humans will prompt it to make stuff 100x better. Our values will mold around this future but I am convinced creative people will continue to prosper in some form.
I think having some sort of identify verification system when you sign up for these generative platforms could be a solution. Possibly with watermarking on the creations as well... The same type of system you'd use for some distributors and banks... that way anything generated that's used commercially gets credited back to the prompter...and licensors can you use watermarking technology to disperse royalties.
I feel like it should be possible for them to encode an invisible audio watermark in the generated audio. Pick some frequencies in the 19kHz range and encode a binary fingerprint that nobody can hear, but you could see it in a spectral analysis. Not sure how post-processing would affect this watermark, but it seems like a reasonable first line defense. I'm sure the people smart enough to code these AI tools can figure it out.
@@AndrewSouthworth There's a better solution than watermarking: force AI companies to upload to something like contentID anything which is generated. So everyone can check whether "your" music is really yours or AI generated.
@@ItWasntAPhase That's entirely a problem of AI companies. If they can't afford to pay artists decently, nobody will upload their music. I don't think AI music can be a business you can make money on. Soon there will be hundreds of AI music generators, since anyone will be able to build them, and they will all converge to the same result, as the technology is the same.
I'm kind of shocked these companies are charging so little for this. Maybe because they figure it will just be open source at some point and anyone will be able to do it, like Midjourney compared to Stable Diffusion? I feel like if all of these services came out the gate at $100/mo for commercial rights it would be better for everyone. 1. It prices out a lot of bottom barrel spam creators 2. It gives more money to the ecosystem for a real royalty model to be created someday 3. The people that will actually use this commercially won't scoff at $100/mo at all, because they're likely already spending 100X that hiring humans to make music. Let's say out of that $100/mo, $70 was allocated for artists who are in the training data. They multiply $70 times the number of paid accounts each month, divide it by the number of songs in the training data and pay it out on a per song basis. Not sure how the economics of that play out but i'd like to know.
A very strange thing to me is that AI right now is very bad at hip hop. Like it will make hip hop but it sounds like it's from the 90s how the verses are structured and it will spit out corny lyrics. I think hip hop has an inherent advantage because it's more personal and based off unique experiences, at least not the generic kind. I know it will eventually close the gap but it makes me proud to be a fan of hip hop. For reference I've used ChatGPT, Claude, Suno and Udio.
I thought the same thing so I told my brother and he put me up on Red Roaring Lion on SoundCloud.. He is using Suno for Ai Hip Hop and I must say he really is trying to master it..Judge for yourself.. I followed and got on train early..lol
@@AskDrLinqShorts not sure how it works. I don't even make money on my own music so not really worried about that. Just think it's crazy it can make music I actually like
@@AskDrLinqShortsAccording to Udio you can use the music you generate however you wish but you cannot sell it. Doesn’t say anything about monetizing it on YT or TT though that I know of. Many are doing that already
Artists that use AI will still need to understand what sounds good that the AI comes up with. Like if you write 100 songs, how do you know which one is the best? If your AI comes up with 1,000 songs, which one is the best? Or can you possibly release or use all of them? Big questions
I'm a creative person, and I for one have been having fun with Suno. I don't think what it produces is something that anyone would mistake for a polished studio production from your favorite band. And if you use it for any period of time, you start to notice a repetition of melodies and vocal performances. But for my purposes -- I see Morrissey from The Smiths walking around Manchester with a wild owl that starts attacking people -- I can go into Suno and write out lyrics and produce a song that makes me laugh. And living with chronic pain, I need as many laughs as I can get.
I wouldn't worry about how these models are trained, when AR Cloud matches and blocks anything too close to something else anyway. You would never know if your music is trained on one of these models if the output is up to par legally. Just as you'd never know if someone took inspiration from your music, so long as they made something original enough to literally be an original work in it's own right. People are getting way too upset over this aspect when there are literally a ton of measures in place legally over similarities in music. When I first started making music 5 years ago, I started by covering songs in whatever chord progression and key I wanted. I used a modified melody. Then re wrote the lyrics entirely. By the time I tracked the song, and produced it in my style, with my lyrics, and my melody, in a new tempo and key, you'd never have any idea where it started. If you could, then the issue is with my interpretation of the song. It's too similar. The same applies for Ai music outputs. If it too closely resembles a song, the output is fraudulent. Just as an "original" would be in such a case. If you don't want music inspired by and melodically and productionally pulled from other existing works, then you simply don't want new music. It's been this way before computers existed. Legally and creatively. We already have laws for the similarities of music. If you want laws over how music inspires and creates new music, then it should apply to real musicians too. Never have I ever seen anyone ask for an upload checkbox to say "Allow other artists to use my track as inspiration for their track" [ ] Doesn't matter how it was rendered, all .wav files are subject to these existing copyright laws, and these Ai gens already specify in their TOS you can't copyright claim other music based off your generations. They do not grant full master ownership for these very reasons.
Is there an option in these tools to plug in your entire music catalog and have it write one of "your" songs? Not that I would release it, but it would be a fun way to see insights on your own songwriting/production and understand the workings of the AI.
I've done this. It's not easy right now. You have to 1) have a lot of songs made 2) upload them all to the service whether it be chatgpt or claude, upload the lyrics, and audio if you can, then write your biography, your musical influence, your accomplishment, your goals etc. Anything you can think of to add more context to the music. Then you can have it generate albums, songs, lyrics, ideas based on all the info you fed it. It takes a long time, but it's fun and rewarding. You can use it to ideate or use it to see where you belong amongst current artists in the industry aka what type of music you make. The tools are as useful as your creativity can take them. Good luck. Lmk how you did.
I wouldn't do that without closely scrutinizing the terms and conditions of whatever ai you consider that with, as I recently looked at one that had small print that using their service meant you gave them non-exclusive terms to use your material to both train their ai, as well as use to promote their service, and so on, so effectively giving them permission to replicate you, learn you, and then give your vibe style to their subscribers, make new versions of your music etc without any compensation or ability to stop them, period. This is dangerous stuff, and artists need to step back, think, look over the terms, or they could be giving permission to be replicated, one they will never be able to undo once fed to any ai company and it gets out there into joe public's home computer and any bittorrent blockchain, etc. If it evolves to a standalone downloadable software not tied into any cloud, then that would potentially be only safeground to experiment with that, unless you don't mind getting replicated without any further compensation for producing 'new' versions of your music, and majorly without your even ever knowing of those future generations.
Good points! Imagine that a few years ago a singer-songwriter releases an album, but then gets visited by music business lawyers who want to see his/her spotify history to see by which artist the songwriter and his album was inspired and influenced by, and then force the songwriter give (a part of) the royalties to the artist he/she has been listening to. That would be crazy, right? I'm waiting for an A.I. Chris Lord-Alge app, that mixes my music exactly like CLA would do. (They should call the app Chris Lord A.I.lge, lol.) One year ago this would seem a futuristic fantasy, now I believe it something like that could happen within in 1 - 2 years. Just like A.I. session singers have appeared. (Audimee, is amazing.)
The mixing AI idea would be amazing. Izotope has been working on stuff like this for a while, but it's not fully there yet. I want a tool I can upload stems, and within a few minutes have a badass mix and master just as good as a pro. It's also a lot less ethically questionable than the music generation, and it's a tool that enables indie artists to have access to world class mixing and mastering for dirt cheap.
Damn. Better than I expected. One more reason that human musicians should emphasize originality and the human elements of their music. There is no training data for originality.
AI shouldn't be used exclusively. There has been MORE THAN ENOUGH "song writing material" out here for inspiration. Artists used to talk about THEIR LIVES. The last fifteen years have been EVENTFUL as all get out. 2020 was like many decades in one year. SURELY SOMEONE can chose something to write bout from that?
People mostly see it as a bad thing if you think in a materialistic way. But once you look at it in a very complex way you'll realize it's got some bad but way more good
seems genre dependant anyone can program beats like this having mastery over an instrument is far more complex. It's like car autopilot we think self driving cars are coming but it gets exponentially more difficult.
AI Music cool. Embarking on my musical journey in the 1970s, I found joy in creating and playing original music with friends, a passion that distinctly set me apart as I rarely ventured into playing covers. As the late 80s approached, my curiosity was piqued by the burgeoning scene of early techno music, a fascination that I explored through the medium of computers. This exploration continued into the 90s, a decade characterized by numerous jam sessions with friends. It was also during this time that I began to engage more publicly with my music, participating in open mic events and seizing opportunities to perform at various street events in town. Despite these ventures, my ultimate aspiration has always been to craft the perfect album. My encounter with AI in music production has been transformative, offering an experience unparalleled to any other. It feels akin to entering a studio as a songwriter, where I am met with an extensive array of musicians and groups ready to bring my compositions to life. This process allows me to meticulously review each rendition, making decisions on whether they align with my vision for the project. This rigorous selection process is applied to all 16 songs, ensuring that each piece not only resonates with my artistic intent but also contributes to the cohesive identity of the album. ruclips.net/video/3Gq7nrG9UPY/видео.html
Andrew I think it's time you think about what music you really love making, no matter the audience, and just make some real art. As good as generative AI is, in all of its forms, it has that zombie, mean-regression feeling. But a lot of intentional music does as well. I just saw a famous but underground DJ the other week and it's like, the AI has no way, yet, to stumble to what makes his productions but also his live performance and ordering of music into a singular artistic statement. The true art will survive. You're a vocalist with a eurorack, f_ the algorithms lets hear Andrew Southworth in his glory.
I've been using it for a week.the stuff I've got it doing is wild.done a 30mins mix. Wow.ive got loads of years making music with my gear and it won't be long before I get bored of AI.
I like the opting-in idea where artists can get paid a percentage for useage. My publishers are currently trying different ways to work with these types of companies to monetize them some how, but it's all in suspense of what kind of outcome will happen for the future of artists. Hopefully, we get T Swift and Metallica to fight for equal rights for artists again if it really becomes a crisis.
I agree with others saying indie artists will be hurt. There is going to be just to much 'noise' out there from this. The ability for persons to create their 'own' music instantly will tie up and clog the world with high-level engaging. People are linear and cannot do more than singular tasks and so I foresee there will be a small staple of industry hand-picked 'popular' human artists, while the rest will be replaced by ai. Already yt seeks to try to get labels to agree to a one-time payment for music catalogs in order to 'train' ai. What I see is this training will enable ai to replace real human artists with the goal of places like yt to no longer have to pay royalties. Without income, many artists cannot sustain themselves. But on a darker note, years later I see real human artists 'disappearing' yet their ai counterparts continuing on, and then a future where ai upped to lifelike hologram and a future world where people can pay/order their fav ai artists into their homes/lives with just monthly flat fee subscriptions as hologram versions, so effectively the music industry for humans is lifeless, just the way streaming has made it majorly unprofitable for majority of artists to even break even, this ai tech will go a step higher to being their own personal 'artists' or bringing their own favorite artists in ai form to them. Already I see people imitating artists in ai characters they are not supposed to do, but people break rules all the time, and this will bypass real artists and make it a joke to be a composer, a musician, when everybody can now be with just a few entries of words to ai already trained on high amounts of illegally gained content they are just not going to give up. It's going to be more and more a scene where we have to pay people to listen to our music, what a joke. To be a music artist now borders on being a vanity book publisher just from the poor streaming payouts and the way platforms are already inundated by tons of artists. A decade ago it wasn't that way. What a difference 10 years makes. Hate to think what 10 years forward is going to mean for new indie artists, such a term may even be past tense, as everyone may then consider themself an indie artist with the right ai subscription.
It still sounds trash. As artists, we see potential in what it generates and that’s why we’re spooked, but it still needs to be brought to life in the studio. Quality with this technology is an S curve.. it will quickly assimilate to a certain quality, but that last 5% will be exponentially difficult for an AI to reach end-to-end. The vast majority of listeners listen to the top .01% of artists - so quality is everything. This technology certainly will add a lot of noise, but the spam will still be trash quality and ultimately lose. Look at blogs, it’s very easy to distinguish AI content from real value content. When a consumer pays attention to what they’re actually consuming, it’s so obvious. And once you’re exposed for that, your reputation takes a massive hit.
Hopefully this content will get saturated to the point where people will get tired of it and call it out for what it is. Low effort content. If you are using AI to facilitate your art... That's cool... If you are using it as a tool to enhance the time, effort, and creativity that you are putting into your creation, then that's fine... But if you're simply going to a website and entering in some prompts, and then spitting out a song that was generated by ai, you shouldn't expect to be monetized. These types of channels should be demonetized. RUclips is about to require content creators to state content was "synthetically generated" but how will they enforce that? I have already come across these comedy parody RUclips channels where they are disingenuous and straight up lie to their subscribers. They literally say that these songs were created by musicians. Or that they found some old record... And people that are not as informed are believing this. Something needs to happen.
I've written my own lyrics and used AI to generate the music and vocals. It can be extremely frustrating because I'm, currently, at the mercy of the AI for the sound and the melodies. But, I finally, after 2 hours, was able to hear a cut that was very close to my idea. Unfortunately, after all that I wanted to change a couple of lyrics, add new verses, but I couldn't. I'm sure those sort of changes will be available at some point, but it can be useful for someone like me, who's had ideas, but can't play the music or afford to pay someone to create it.
I want to feed my demo to AI telling to keep certain things (lyrics, melody, main harmony etc) and prompt to alter style and sounds.... kind of fast producing. Could focus more on composing and let AI to produce different genres....and with different singers (Elvis, Chester, etc) As owner of the song I would be safe and possible to reach new audiences. Graphic designers and producers....basically all those making the endresult shine are propably the loosers here. Sad but true. How long do I need to wait? 6 months is my prediction. Ps. Still learning to use all plug-ins and stuff despite knowing soon to be almost useless knowledge. But hey....try to do very specific image with Midjourney. Impossible. Same will happen with audio generating so keep you producer skills tuned still😊 And thanx Andrew for this great channel. Makes world a fraction better.
I'd think the better this gets, the less opportunity/value of real artists. However, I think it could still be used as a tool. It will just become a competitiom of who can manipulate AI best to get the best song out of it. I'm surprised by how good it is and have a lot of fun, however I still value human-made music over this. Nothing beats natural-creativity that's expressed through music.
Another important point, is about music composition. I can understand if AI become insane producers in term of quality etc. But when it comes to melodies, there is no mathematic formula to make a great melody with emotions, it's all about feeling, and AI don't have that. So there is also hope here, i hope those tools could become much more customizable and tweakable.
I don’t know about that in re: AI melodies not having soul. I mean, yes, I guess you could argue it’s not literal soul, but that 1940s style female vocal track Andrew plays in this video sounded like it had soul to me. If you walked into a cocktail bar and heard that playing, I don’t think you would think it wasn’t a real person with a real soul who had made it. Another argument is that AI music is an amalgamation of many pre-existing souls combined- real people who felt real things and made real music.
Hey Vanessa, thank you for your reply, i agree with you that this example is pretty good! Personally i would say that the vocal melody is ok, not super great but it's indeed impressive! BUT, to me, it's the voice that is making it much more emotional, it's really impressive, but the thing is this voice come from someone else voice, it's not a 100% synthetic artificial voice, it come from human data, and i am pretty sure that the first problems that will come related to AI music will come from the voices, artists will hear their voice in some AI songs, it will be a lot of sues i believe, (there is already many examples, and it's already happening). So, yes, there is emotions here i totally agree with you, but it's like cheating, cause it come from a real person voice.
But if you do the test, you play just on piano the SunoAI generation, you translate them just with piano, you'll hear that many often there is no senses in the melody, it's not what a human would have done (usually). BUT, i believe that through a ton of quantity generation, you can with LUCK (and no skills) have a beautiful composition
The thing is, it knows what melodies humans find emotional or powerful, so it can generate new melodies that also feel the same. You can genuinely make tracks that feel like they have so many feelings behind them, pain, love, happiness etc - except it's just an algorithm engineered to make you feel those things. It seems like at the moment it's actually better at capturing the melodies than the production quality. Apparently on Udio's roadmap they're planning on allowing people to download the stems for their tracks, which is a huge leg up in making these sound great or using them in more useful settings. Apparently you'll also be able to use it to generate samples and sound effects, which is also super useful for a myriad of reasons.
Yeah from my tests, the average person can't tell the difference between human made and AI music. We might be able to tell, but we're all actual music creators. We buy fancy speakers to mix our music when 90% of consumers listen on crappy headphones or their phone speakers. I don't think AI music will ever be better than the best humans in the world, but its already making better music than many humans in the world. I think humans will start finding new avenues of music that AI can't do well and dive into those areas. Basically cookie-cutter music will now be mass producible with AI so humans will be forced to get more creative and innovative to stay relevant. I think human artists will also have to be even more open about their process, and share even more of their personality. Because the process and their personality will be what makes them shine and stand out for the sea of AI music. In some ways I feel it could create a renaissance for music.
My music is all AI done.... and I'm trying to become one of the first wave of AI assisted music artists to reach a 6 figure income from their music. So I'm doing the climb one step at a time with the music to try and get there. I don't understand people against AI. As an artist the important thing is making use of what we have to make new art. I'll leave all the questions of behind the scenes copyright law to lawyers. But right now everything 100% legal and legit.
You made a great point. The only way AI music can be legal is to tie it to distributors, and AI companies will have to offer yet other platforms that we upload music to, if we want. That's the only legal solution. That's what will happen, as it's not true that AI companies do what humans do: in order for a human to be "inspired", they have first to consume legally the music, so pay the copyright holder for a license. AI companies have to pay as well. That being said, I don't think it would be smart for labels like Universal to license their music to tech companies, unless they have a partnership with them. Suno is just 12 people, so literally any label can make its own AI and earn money exclusively. Once a technology reaches mainstream, it costs nothing to replicate. That's the self-defeating nature of AI: since the machine does all the learning, the human factor counts nothing. So again these are all the premises for a new AISpotify, with which labels will be partnering.
I don't think Suno et al. drastically dilutes the music pool anytime soon - you still need to take these ideas into a DAW and re-record and re-produce/program everything. In other words, you still need to know what you are doing to get it to the right level of quality. I also do not see "average" people getting heavily into Suno anytime soon either - even if the quality is there, you still need to hop on the app and enter prompts. It requires work. Humans are wired for efficiency/laziness and going on Spotify and having a playlist made for you is still easier than that. The better use case for Suno right now is as a writing tool for existing musicians/producers. In a year or more though? Who the hell knows.
It is also being used by content creators for intro/outro/background music. I can see game companies use it, smaller tv productions and maybe even ad agencies to come up with ideas
Honestly, outside of the scale and speed of it, how is A.I. learning from music any different than humans doing the same thing? I've I've never played Jazz but heard it and liked it, then I went out and listened to every Jazz song I could find and allowed that to teach me how to write a Jazz song, would that not be the same thing? Not trying to be sarcastic, just I honestly don't see the difference.
In defense of AI inspired music I would like to remind everyone that every piece of music that has ever been created was inspired by existing music. When it comes to AI generated music I have to assume these completed pieces are being inspired originally by a few lines of text, followed by a multitude of bits and bytes the computer puts together to create a song. So leave it alone because what we as humans will end up with is a better musical product that still has a human element attached to it.
Every time AI shows up in a new space...the people in that space make the same mistake. They conflate the "end product" with the "AI model". They are act as if they are the same. They are not. The AI models are fair use. Why are they fair use. Because they are transformative. That is...they don't resemble the original in any way...they don't have the same use...and they don't compete in the same market. It's transformative. Transformative use is fair use and fair use....well...you don't HAVE to "ask permission" to use someone else's work. The math models don't resemble the original works in any way. So they are fair use. Now one could say, "Yeah but the people USING these software are "ripping off" the original artists." Well...that would be a case that would have to be made against every single person USING the software. You'd have to be able to convince a jury that a udio user had made a "copy" of your song. But you can't go after udio itself because they made a tool that someone MIGHT use to copy your song. That would be like trying to sue Adobe because they make Photoshop and bad actors use Photoshop to violate copyright all the time. In countless ways. Until you can come up with a reasonable line of thought on why Adobe should be able to be sue because of the ways in which "some" people use their product...you cannot make a reasonable case that udio or OpenAI or any of the rest CAN. (And please don't start with the "but they're making money with it so it's not fair use" misinformation. There is nothing in the fair use doctrine that says you cannot make money when you use material in a fair use way. Otherwise all those satire cartoonists in The Atlantic that poke fun of major copyright IPs like Star Wars and Marvel would never be able to be paid for their art. But we know they ARE paid. Even though their art falls into the fair use category. Why? Because you can get paid for your fair use work.)
The piece that I argue should be controlled is not the usage of the output, but the data used to train the model. Everyone should have the ability to choose whether they want their IP used in a training model that can be used to output new creations. Granted, you can't choose if someone makes satire of your work. But these generative AI systems are a new beast that I don't think should have the same 'rights' as a human. I feel there has to be a legal discussion between all the various stakeholders in the industry behind what rights and safeguards we want these tools to have. At the moment it's the wild west and it will take governments years to decide anything, but that discussion needs to happen to decide what is best for everyone involved. Maybe the answer is as simple as all these tools must watermark the output in someway so that it's easy for everyone to tell if the audio was AI generated. Then platforms display all AI work as Ai generated, and the consumer can decide for themselves if they want to support or skip listening to AI generated music. Perhaps AI generated works get paid less due to the ease of creation, and it prevents royalties being taken away from actual human artists. Or possibly the answer is more complex, and we decide all training must be on an opt-in basis only. And people can negotiate how much they want to be compensated for having their work be used in an AI model.
To be honest I think ai music sounds as artificial as AI images. I made a lot of songs that sound great but after a while I started to get nauseous and my ears hurt after listening to them. I don’t know which frequencies they have but they male me sick
AI-generated music has made impressive strides over the past six months (since this video was published), but it still won't replace real musicians. Live music offers a unique, irreplaceable experience that goes beyond just the sound-it's about connection, energy, and atmosphere. That said, for jingle writers and stock music producers, the story is different. AI has essentially made their roles obsolete. But let's face it: nobody's making real money from streaming anyway, whether the music is created by humans or AI.
Imagine you subscribe to a service like the one you are describing. How can you prove that your song(s) has been used to create another song and therefore you deserve to be paid?
To me, these AI-generated songs sound soul-less, and yes like they've been done before. I'm curious if anyone has analyzed the sound quality, loudness, width - essentially the release-readiness of this crap?
I've gotten it to generate really good sounding math rock songs that are way more moody and even have fascinating song structures. It's horrifyingly good. To clarify, I'm not using it in my work at all - it was just an exploration.
@@nova_harbor good to hear it was just an exploration 🖤I believe AI could potentially be a good thing for somewhat more "creatively daring" musicians (e.g those artists not trying to sound like someone else) which could see a lift in more experimental and niche music. That would be cool.
having spoken to these AI "filters", no, they make zero decisions, they type it in, grin because monkey grin at funny machine, then dance when the music comes on and say they are "producers", so, they are less than knob twisters.
They're definitely not 'producers' and most of them are just acting like monkeys in front of a computer. However there are also people who spend a lot of time perfecting their prompts, generating dozens of versions to find the perfect one, using other software tools to convert them to stems and then mixing / mastering the stems while adding new layers to improve the track. Just like actual artists, there are different levels of work people put in. Some artists buy a beat online, lazily freestyle rap over it and then throw it online. But other artists make everything from scratch, obsess over the writing for weeks and properly mix and master the song.
Ai music, a lot like Ai in other industries, will not kill music creation. People will always want music from people they can connect with. Problem is AI music will flood the market and devalue artists. Also the landscape of music in commercial uses like games, commercials and content will drastically change
I don't think there's anything wrong with it. A lot of artists try to sound exactly like other artists. It's called inspiration. It's not stealing samples from anyone. Samples and presets are different things.
Okay i am way past the "Fuck, i am angry." I more into, how do i make this work? I mean what do you as an creator use this and give your spice to it? It's got all the instruments and songs. Do you sing with it or what?
Great idea but keep in mind that the tos says you can use the music however you wish but you are not allowed to sell it. So the songs you create can be monetized to stream but now sold as tracks or an album
yeah great idea.. let's all sing with it.. bright future ahead of us everyone! nah fk it, im not laycing my music with it, who knows what kinda subliminal messages is the algorithm putting in 😬🤔
You talk about companies having a "right to train" their A.I. models on pieces of music. When someone goes to school for music, are they only allowed to utilize music that the school has acquired some sort of approval to train their students on? Is a teacher not allowed to make reference to a song in order to teach music theory or composition if they don't have approval from a record label? What is a "right to train"?
AI cannot be 'inspired' by something, because it is not capable of independent thought or feelings. It's not taking inspiration from songs to create something completely new. What it's creating is an amalgamation of it's reference data. So in that case, it could be argued that musicians DO deserve royalties for their music being used in training data. If you're building a house, unless you make everything yourself, you have to go out and pay for the bricks, the glass, etc. You don't just get given the material for free.
I just made a collab with AI. It's a video up on RUclips now. I wrote part of the lyrics, then the AI wrote the middle stanzas. The song I did before that was done with a plugin where the rave instruments (that I don't have) made an entire song for me if I just pushed the buttons in a row every 8 beats. What both were doing, were playing popular note progressions, ones that historically sound good, and that we equate with good music, what we want to listen to, with great tension & tonic motifs. These riffs are in a lot of techno songs. At first, I felt like I was cheating; I'm just a guitar player with some minor piano skills. I love the latest 2 songs I did, but I didn't do them; it is weird.
When I saw the ad for 'Donna' I got excited because I thought I could use it to make references of songs I've written myself. Finding qualified reference singers has always been a thorn in my side because I've either run into singers who only want to work on their own music and are unwilling to help out - even when paid, or I haven't had the amount they ask at the right time. I saw the A.I. app and thought it was the answer. But, all it does is create its own songs. Yuck. Songwriting is my first love. I take pride in stringing words together and creating songs with my own mind. It's like instant grits. We all know how My Cousin Vinny and any self-respecting southerner felt about instant grits. It's just not quite as good. Where's the artistry in an app like this? Where's the soul? This is nothing more than a dream killer. But, as stated in the video, it's here to stay no matter what we do.🤬
As a venue for music I would never book AI music acts. Just out of ethics. If I get the slightest hint that they are just AI based - off you go. Maybe won't take long until there are careers built out of this AI covering of songs but I guess we already have a lot of repetition of the samey old shit. The worst part of this is. It devaluates human creation with its natural flaws and beauty. Cause what these songs need to do is to appeal to people who just want to throw out these working songs. If they would be just amateur or something like that people wouldn't use this. So there is an inherent aesthetic and hearing habit that was put into there as a bias. Be nice and support your local human creators.
The Artist have to go the modern way right now! Write own Songs, create AI Songs. Get Ideas from AI and Remix ur own generated Song. Of course its harder to get into the game now but the listeners are blessed with new nice Music. I think its a win win on both sides!
I think AI Music is really good for those who feel they dont have what it takes to live there dream, im a singer and a lyrisist, i have no band and i intend to sing on my track that i make, i just dont have a band, im nearly 40 so finding bands im into in my town at my age is hard, NU-metal with pop vibes
I feel like this is going to ruin singing artist or producer and here comes the prompt artist replacing producers all kinds of stuff or maybe we just have to raise are prices and put value in our music and lower prices in ai
100% I'm looking to save money by getting AI to sing hooks for me instead of getting a feature artist. I do everything in my production process except I can't sing. I make hip hop music so every now and then I'd need a singer.
There is a logical basis to what you say about how the artificial intelligence was trained, but the only thing that ultimately differentiates the artificial intelligence as a music producer and the human as a producer is that the human chose and listened to 10-100-10000 songs from which it (possibly) drew inspiration (imagination - influence) and the AI was FEEDED (so it also listened, but without conscious participation) 10-100-10000 songs from which it drew inspiration (imagination - influence) and created something new. I will take you to the case of theft of intellectual property of the invention of car wipers, where the inventor, when accused of having used EXISTING electrical devices in his invention (so "it should not have been accepted as HIS invention", but as a different use of existing inventions that gave the effect it gave), he answered that the words used by poets in their poems are also given, but this does not make one poem a product of theft from another on the grounds that the same words have been used in one and the other. Likewise, here too, we all know that when it comes to music, there is no parthenogenesis. We have all been influenced by something (either consciously or subconsciously). I find nothing wrong with the exact same thing happening with artificial intelligence, even if in its case, the influence comes through human intervention.
Generative AI models are trained on copyright-protected data - is that legal, is it sampling, scanning, theft. How can i protect my art. How can you say no to your art be scanned by ai???
That might be true, but you can still commercially exploit an AI track. Also as far as I know that's only for registering copyright, which most artists don't actually do anyways.
Hey Andrew thanks for the comment. I’ve enjoyed your content for along time. Yes, that is true. Also as part of the Udio agreement which you sign when you register you have to give them credit if you’ve used their platform to generate a song.
Over 6 million songs have been generated by Udio in the last week. If those songs are uploaded to streaming services it further dilutes the streaming payments to artist that create without AI.
Yeah for Udio you have to give credit, not sure if their new pro plans remove that requirement. With Suno you do not have to give credit and as far as they're concerned, if you generate the audio you own it fully with no strings attached.
This is scary but I've tried it. I play guitar and keyboard and sing. I'm gonna use this as a tool. It helps me with ideas. If my AI songs become hits I will make sure they know I used AI, I've used chatGPT to create lyrics but I change them more to my style.
But you see how little time it took him to type in those words to have the AI generate that? It isn't right for people to be monetized for putting in no effort meanwhile RUclips content creators are putting in tons of time, effort and creativity.
You're not learning how to do something and you're not creating anything. You're placing an order with a copy machine that imitates the creations of others to manufacture you a product.
Composers and music producers need to get the courts to force AI companies to give each of their artificial creations a registration number. This number must be saved permanently so that the AI song remains identifiable as such forever. And therefore NO copyrights can be claimed!
You talk about companies having a "right to train" their A.I. models on pieces of music? When someone goes to school for music, are they only allowed to utilize music that the school has acquired some sort of approval to train their students on? Is a teacher not allowed to make reference to a song they in order to teach music theory or composition if they don't have approval from a record label? What is a "right to train"?
Treating the AI like a human makes sense in certain scenarios, but they should absolutely have to get permission for training. No question about it. Huge difference between a music student and a machine that can crank out thousands of songs per day. This isn’t a sentient creature we’re talking about, it’s an algorithm made for profit.
@@AndrewSouthworth So what if it's not a sentient creature? It simply does what a human can do on a super human level. And even if it is made for profit, that's what a human does. A human looking to enter the music industry trains itself and then produces music for profit. Often times these small-time music producers don't hardly know any music theory. They just know how to create a four-chords song that sounds like a popular artist, then get revenue from it because the masses have been trained to enjoy repeated four-chords songs. Or what about music libraries. Is making simpleton, stock music for advertising really that much of a dream job? Suno and Udio are not stealing the music any more than a human is when it's analyzing a chord structure or riff or instrumentation by ear. They're not distributing it or claiming to own it. All that can really be said about these companies is "it's not fair". Personally, I think it's already a crime that the music industry has made as much money as it has off of such formulaic drivel. That's my opinion at least.
In my opinion, if the company is going to release a commercial product they should have to get permission from the artists to use their music for training purposes. When a human learns music and get's inspired by artists, they actually pay for music. They stream it, go to shows, buy tab books, take lessons etc. The AI does not compensate the artists like a real human would while learning to make their own music. At the very least, they should be required to at least purchase every song in their training data as a digital download.
@@AndrewSouthworth Okay I can see that. They pay for the music, like as a digital download. And then maybe they sell training data perhaps bundled by genre. And any music a person generates from their purchased training data they have full copyright for. As of right now, Udio is not selling their A.I. service. And I don't think users have copyright for the music they generate. So maybe if it stays this way it remains a free service, but giving the users copyright over generated music would require the purchasing of training data. That makes sense to me.
Training is not how you think, also is possible training it self. Even training on artists music is the same as if someone just heard the music Notting more
I can see this as a threat to those who produce cookie cutter pop and rock, but do you ever see it get to the point where it can make a genuine issue Stairway to Heaven, Paradise City, Bohemian Rhapsody or even My Dark Twisted Fantasy? Certainly don't foresee it able to create a Brandenburg Concerto or Pastoral Symphony. Generic music yes, monumental works of musical art, not seeing it yet.
The drum sounds alone are of the same quality as in the 1970s. Guitars are difficult to track and are of poor quality. You can currently only use this stuff as a suggestion and replay it. But then you're very poor if you can't think of a better song.
I don't think so. People thought recorded music would destroy the music industry, then radio, then physical media, then synthesizers, then autotune, then digital downloads, then streaming etc. The people that care find a way to keep doing what they want to do, and the business folk keep finding ways to make money doing it.
@@AndrewSouthworth yea but AI is something else mate. Radiomusic and all other music came from humans and now a robot can do such melodies. So bye be music industry as we know it because any human can make kinda good sounding music with one mouseclick. time to focus only on live performances with instrumental and vocals.
@@rolandvonutopia It is making music trained on other music. Human music and creativity had to exist for the AI to function. If you make boring cookie cutter music, yeah, AI will probably replace you. If you make interesting and creative music, AI can't touch you. AI is not creative.
I would say - don't let the companies scrape all the songs. As a legislator I would get into their data system if I have authority and check whether they used copyrighted data. Which they definitely did because the open songs that are on the net are not sufficient enough to create such vast amount of styles and proficiency. Wonder who will win in the case. Same as the visual art. I have the suspicion that with legislation there will be a lot of court cases. I am against the creation of these songs as a legitimate art form. They are not art. If you as a human are not putting in more than 50% of your own creativity then this is not art.
![Blox Fruits Dragon Rework Update [Full Stream]](http://i.ytimg.com/vi/EqDAp8udhm0/mqdefault.jpg)
Before you comment, let me clarify my viewpoints on this:
1. This technology is going to exist no matter what we do, we need to learn how it works and figure out how tools like this can help us as music creators.
2. Someone needs to create a system that properly compensates artists who are used in training data, and it needs to be opt-in only. Nobody should be used as training data without their permission.
3. Governments need to act fast to impose laws for how music is used in training data. Not everyone will follow these laws, but they should still exist.
4. This isn't going to replace artists. Don't panic. Fans want to connect with real artists doing cool things and people mass uploading AI music can't replace that (at least not soon). This will however likely screw over a lot of artists making more templated music such as lofi, meditation music, corporate background music etc.
Please read the terms of service or check out the Top Attorney episode on the TOS. You can’t copyright or have sole rights to the masters. That makes this tuff useless.
With other AI´s that create videos and photos you can create a whole AI person, we already see that with AI influencers. Now the only thing you need is a hologram of that person and you can do live shows, or virtual shows (apple vision etc.).
I belive that we will see a lot of those "artists" in the near future (5-10 years). Think about the budget cuts for major labels, they don't have to share with an artist anymore. Hire a few designers and prompt writers + a producer to touch up the generated music and you are good to go.
Governements are no longer the answer because they are in bed with the big industries, and they have been for a while. What will happen is already decided, but it must seem to come from us. We need to choose it, even if it would not normally be our choice. For that to happen, we (the people) are going to suffer it first. It’s a long con false flag operation.
Why do governments agree with this? Because they have been fed a certain apocalyptic narrative that shows what will happen if they don’t. (See Daniel Schmachtenberger’s explanation of this issue for more info.)
Partly agree. There are also many good producers who work in the background but don't connect with fans whose music is not "templated". Their economic existence is definitely under threat.
But there is a point about people, they are always something new, and AI is not able to innovate, to create something original, these sounds are generated from models created with songs by other artists, that is, generic music, I believe that, many people will lose space for sure but original music will be even more requested.
New small artists are the ones drawing the short straw. There's going to be countless AI songs out there to compete against, and it won't even matter if you use AI or not yourself. You're still going to get swamped against the vast amount of new tracks out there. The outcome for new artists will be more PR, more social media content, more non-music stuff to get yourself heard.
Established artists will be affected too of course as there is only so much ear-time of listeners out there, but I'll guess they'll be affected a lot less. Thanks for your thoughts.
Yes, agree. think that as the market becomes oversaturated with AI, real things like live concert performances / live stream performances will become more valuable.
the out come will be that youre gonna make more money than ever, doing live shows
it just means artists are going to have to work. you see how lazy Beyoncé has become with her latest "hits".
@@GhostWriter_Music Yup. Musicians will have to work much harder making better music, more content. My head is already spinning. I might need to see a priest about that.
@@GoranBackmanMusic no longer will a studio or artist be able to just throw money at a problem and it get them a bunch of fans and profit, because that will be expensive.
Ive been in a touring indie band from 2013-2016, and now a full time music producer working for trippie redd, kehlani, machine gun kelly and more, and I have to say Andrew nails it when he says this tech is both amazing but also terrifying. I don’t think we have any idea how this is going to change the landscape within 3 years let alone 10!!! I don’t think real music is going anywhere anytime soon, but I 100% agree it’s good to keep an eye on this technology and educate ourselves. Great video as always.
I just came across this video and found it to be very informative! Thank you! I have never written any lyrics or songs in my life before finding a RUclips video about SUNO. I have been using SUNO AI since September 6, 2024, and began writing lyrics using bible verses or things I have written in the past, and creating music videos. I truly believe SUNO AI is giving people the opportunity to be creative and it doesn't cost a lot of money to get their creative products out there. Thanks again for a very informative video!
If they owned the rights to the training data they wouldn't be so silent about it.
That's a good point. Meta released an open source music generative AI tool a couple years ago, and they were very vocal about the fact they trained it on 100% properly licensed music.
It's an awesome marketing angle to be able to say not only is your AI tool amazing, but it's the most ethical one on the market.
Didnt the big labels go after them regarding this?
Been telling people this the whole time. People seem to think ai has no soul so it won't take off.. they are clueless and don't see the writing on the wall. People will generate the songs and then add the soul to it when they modify it. The market is already over saturated and it's going to get worse with ai. Big artists will use it and now they don't have to share publishing with the people who were writing their songs cause they can now create their own songs... it's going to be crazy.
I am strictly a lyricist. Not the best, but I write a pretty decent song. Can’t play, can’t sing. Been writing for about 15 years. Prior to AI, I would finalize my lyrics, sing my song into my voice recorder, send it to some Nashville studio, pay them anywhere from $300-$1000 and wait 7-14 days to wait on mymasterpiece. Was always happy with ny final result, but there was always something you wish was different. And over the last 15 years pitching my songs, I’ve had to endure ego after ego, gatekeeper after gatekeeper. Disputes, you name it.
3 months ago I stumbled on Suno and paid for the pro service. Inputted my lyrics, made my prompts and let’s just say what they generated were literally 100 times than I even envisioned for the song. So now I pay a small monthly fee, and get literal masterpieces in my living room in seconds after completing my song. No egos. No drama, No paying $750 to wait two weeks from some premadonna who ghosts you or tells you what your song needs not the other way around. After a few days of what I stumbled upon and seeing the results I was so inspired I proceeded to write probably 100 songs in a short amount of time. Songs about anything and everything, with my own lyrics. Beautiful stuff. Wrote songs for every person who’s ever been in my life, and moved them and their family to tears. Now have a significant music industry vet in my corner, getting interest in my songs that contain my own lyrics. The lives that these songs I’ve created with Suno’s help have literally moved so many people, not just myself.
But everyone wants to tell me ai is not good and the music isn’t good. I’ve had songs placed before when I worked with humans, had my share of compliments.Songs picked up by the NFL, NHL, etc. And my songs were pretty damn good back then. Now they are mind blowing and now for the first time I have a fighting chance to compete with the big boys. Right now I have a 30 year Ivy League entertainment attorney submitting my songs to A-list artists. I use my songs as work tapes. Currently have songs out 20 major artists. All 20 might pass and I might never get a cut an again. But I tell you I have about 100 of my songs on my playlist and they are the sound track to my life. With that said, people who say AI is soulless and the songs suck and, just aren’t paying attention or some musical dinosaur felling like they are about to be extinct. I ❤ ai music. Who the heck doesn’t want a music studio in their home that doesn’t ghost you, lock you out, send you duds and keep your money. I get what I want and I get it when I want it!
I made AI album a year ago. Been saying for a while producers and artists gotta curate a sound and build and market that. I bet Spotify will eventually generate AI music that matches YOUR algorithmic tastes. And the royalties will go to them 100%
I'll add on top of this. Soon artists won't need to pay producers when they can generate a million type beats until one is close enough. They will own 100% of it. Producers. Adapt now. Invest in artists not beat sales.
100% agree there could be a day Spotify adds an AI music feature that generates infinite AI songs that adjust based on what you listen to most, and what you skip the most. The good songs get turned into permanent tracks that they could then recommend on other playlists.
Honestly this is probably the most dystopian situation I can imagine, so I really hope this doesn't happen. I feel it all relies on if the average consumer enjoys the tracks or if they reject the idea of it. A lot of artists would push back against Spotify doing this too.
But from Spotify's perspective, artists and labels take 70% of their revenue. If they can cut any part of that out it significantly increases their bottom line.
I feel like this particular aspect could be illegal though since they're a platform - just like how Apple isn't allowed to promote Apple Music in their store more than Spotify, despite the fact they own the device and the store. Also just like how Microsoft got in trouble for pushing their own browser. Congress could view Spotify pushing their own internal music as stifling competition from the rest of the industry since they have so much marketshare.
@@AndrewSouthworth I Agree with you on this. Then they will work with the majors to find the hot new young performers to 'play' the songs live.
Something about blasting artificially generated material in on the same level with human created just seems morally wrong to me, music or otherwise. That’s the larger issue with all this I think. Will they allow robots to compete against humans in sports ? I don’t think the training or curating or “knob twisting” is particularly relevant. Michael Jordan playing against a kindergarten basketball team doesn’t add up
I can see this happening. It's very likely we are headed towards a scenario where everyone's life is AI curated in every aspect, as far as entertainment.
One way to slow this down? The AI companies need to be sued, and they need to stop being able to scrape the internet for data. The tech is only relevant so long as it is able to scrape all the current data.
@@AndrewSouthworth Spotify already commissions music from producers, I've met some of these guys personally. They get paid upfront fees (e.g. $1k) to churn out EDM for Spotify's editorial playlists and waive all rights and ownership of the music. These songs have millions of streams and fake artist profiles. AI music is the logical next step for Spotify et al and I fear the average listener won't care where it came from, they'll just listen to a playlist and go about their day. Human made music will be niche - like if you were to go buy organic produce at the farmer's market.
I think you're absolutely right, this has probably existed for at least a decade and the public is just finding out about it. If we have access to this for a $30 dollar subscription, just imagine what they have access to in a million dollar recording studio.
I make AI music, as a hobby (i write my lyrics) i just choose a style and also i edit what i dont like, Music is beautiful no matter how its made, the future is here, the same was said about digital keyboard, with loopers, this is just much more advanced, Nothing takes away other peoples music ever! true artist will always be better, even as an AI Music creator i cant compete with real music
00:00:00 - Introduction to AI-Generated Music and Its Perception
00:00:28 - Creating a Song with AI: Demonstration on Two Platforms
00:01:15 - Song Comparison and Capabilities of AI Music
00:02:37 - Exploring AI-Generated Metal Music
00:03:10 - Understanding the Quality and Limitations of AI Music
00:03:40 - The Ethical Considerations of AI-Generated Music
00:04:41 - Training Data and Copyright Concerns
00:05:16 - The Future of AI Music and Its Impact on Artists
00:06:08 - Potential Solutions for Fair Compensation in AI Music
00:07:02 - Generative Music and Ownership Questions
00:07:36 - The Role of Artists in AI Music Creation
00:08:02 - AI as a Tool in the Creative Workflow
00:08:32 - Conclusion: The Future of AI in the Music Industry
00:08:59 - Final Thoughts and Viewer Engagement
Parody music generated by AI has taken off like crazy, there is an AI artist on spotify with 200k+ monthly listeners with music from suno, and youtube plays in the multiple millions.
What is the artist name and youtube channel?
Yeah what’s the artists name
Obscurist Vinyl
Why should this be monetized? Little to no effort, little to no creativity, hardly any time to generate. Reddit is literally filled with discussions on people very upset that AI content creators on RUclips are trying to grift or get something for nothing.
@@ChrisCeeKayKelley they should be definately monetized, there are efforts.
Oh my lord. It really is insane what it can do. I think the creators in the music industry most affected are going to be the producers or anyone who programs music, be that EDM etc. The winners will be the actual musicians who can play instruments live and are willing to go on the road to do so. Even if you use AI to write the songs for your band, it cant get up and play them for you. You will still have to that. And DJ's? The Ghost producer days are numbered for sure, but DJ's are curators at the end of the day, so regardless of whether they make the music or AI does, they still decide what to put their name to to and play at sets, so don't see a big change there.
Yeah it's kind of like that newish South Park episode, where all the white collar workers don't know how to do anything anymore and so the handy man workers become billionaires because they're willing to do stuff with their hands and learn how to fix stuff haha
I could be wrong, but think with me, everyone is going to have access to AI Music, and it's not able to create 100% original music, so everyone is going to have the same type of music and they're all going to sound generic, and today the artists that emerge is because they have something unique, different, they deliver something innovative, and the AI is not capable of that, so about Ghost, it is possible that for some artists they stood out and a lot of them don't know how to create music, they can turn to Ghost who know how to create something innovative and that makes sense in the project of these artists does it make sense or am I being too optimistic? Because it's impressive what AI Music does, but I've only heard generic sounds so far, I believe they will improve the technical quality of the sound mix, but "creatively" think they don't go beyond that at least for a long time
@@Gutz-po9xf I can promise you that you are not correct. The top tier AI output is absolutely as original and inventive as the very creme de la creme of inventive and original music from the past.
You just need to look a little deeper.
@@AndrewSouthworth great episode and that is exactly how i see the future playing out. BUT upon a foundation of universal basic income; where people can then earn extra credits based off effort / exchanging / deals
Winners are vocalists and lyricists. The copyright office better wake up and stop claiming everything made by AI is free to copy because people will lie about how something was made, and lawsuits will fly if people rip off other's lyrics and also voices and vocal characters. Losers will be generic composers who have already been losing for over a decade anyway. The film business is also imploding. Eventually govt will have to step in, but in the USA, we've got a rising tide of fascist religionists who hate govt support for the arts anyway, so best idea is to move to another country if you want to be a creative.
On average, popular music has already become more polished, less natural, and even somewhat ‘robotic’ in the past 20 years, and people seem to have accepted the often ‘plastic’ feel of it. That alone should be enough to indication that A.I. will definitely replace a lot of musicians. But only those who don’t understand how A.I. works, and also don’t understand humans, will say that it will never be able to replace musicians. The key here is not A.I. though, but humans.
Although we have made scientific and technological progress, as humans we have not truly evolved. We are using the same level of thinking and consciousness that we have been using for at least a couple of centuries now. It is precisely that kind of thinking which has led to these technical advancements, but which has also kept us away from real human progress. Once we start to realize this, society will start to change rapidly and it will choose a different direction. Hopefully, this realization will not come too late.
Very good point. People already accept plastic mediocrity so AI won't bother them. We are far from "Pink Floyd - Shine on crazy Diamond" days
That 40s thing is just an effing joke ... the lyrics don't make any sense and it's just so generic. Gimme a break. It's crap. But so is MacDonalds and plenty people eat that - but they didn't put good restaurants out of business ...
Most casual listeners of music, don't care about how the music is made. No music producers or singers make songs without being inspired by someone else.
Just think how much AI has lifted the bar on vocal skills, musical skills, arrangement and production . The worst you get is some cheesy mess, the best is truly scarey good.
give it a little more time! This is really a problem for all artists! I don't think there's much we can do. The music industry was already crappy after Napster..the death of cds, then Spotify royalties and play revenues..now this? it's just getting too much to bear. I wonder why should we keep trying? it's really serious.
I bet they're already musicians pumping out lyrics or albums really that they don't have the skills to, and then cover those songs live, I mean who can tell? I struggle with a blank paper to write lyrics, come up with the music and putting stuff together..now you can only be a cover artist ..hell..you don't even have to be a musician at all. Im not against technology, but this crosses the line. And I don't believe there's a law that can control this from happening.
They could backup automatically every AI generated music and use the existing means for copyright control scanning..but how about I record the song generated all over on reaper and then upload to CDbaby?
this os "cheat gpt" and no way to know. A band can release 6 albums a year or 2 each month while the honest guy will come up with one, maybe two in a year if they're really inspired. Not fair and quite horrifying for us all.
I honestly think it will be the same as with AI art. People will use it, creators might use it but ultimately it will give more value to human made art. People need that human connection.
I agree. If everyone can make decent art with AI, the people that make amazing art from scratch feel a bit extra special because they didn't NEED to do it from scratch.
In the end human connection is everything exactly my thoughts
I'm trying to see if a large portion of music listeners and creators are going to avoid playing AI created songs, to the extent that they petition spotify or soundcloud to filter out A.i created content so they can only listen to real artists. This could put a big cut on AI stream count and numbers/royalties, but just thinking that most of us never asked for ai music. I'm open to thoughts on that, as I don't know if that would work or how it would play out, I just know I don't want to listen to ai created content... maybe just an internal bias I'm having I can't explain.
@@AndrewSouthworth They actually did need to create it, because AI can not create what a human would create. A friend of mine proved that no combination of prompts could actually match his mind's eye. Creating art is something only sentient beings can do, and AI is not sentient. If I smash my printer, that can be considered art. If my printer falls off the table and smashes because it is shaking too much, that can not be considered art. The difference is in the intention behind the act. And no, entering prompts is not a creative act, as it lacks predictable results. Art must have CREATOR INTENTION behind every detail. This is why we do not consider a pitch to be a movie.
I don't think that simply entering a prompt and getting the output is art either. However as these tools evolve I suspect there will be plugins and tools created that assist artists in creating. For example, programming a simple drum pattern and then using a generative fill feature to humanize and add much more detail to the drums.
When you think about generative Ai tools in this way it's a lot more artistic.
There are also tools being added already that allow users to customize their AI creations using 'inpainting' for audio. This is still not very artistic but it's one step closer to giving the user more control of the final output.
Embrace A.I. It is only going to be bigger. Artists? What artists? They are manufactured now. Get a model, write some lines, and auto tune voice. A.I. does it better. Compensate? That's like Crayola suing you, because you made a great work of art with their crayons. Connect with artists? Why? I just want good music. What about music that doesn't have any vocals? Smooth Jazz. Does it matter who or what made it? Btw, the music industry is trying to own the rights, and create their own A.I. That way THEY will be able to make music in the future and not pay a single human. They will own it all.
I downloaded Synthesizer V to help me complete songs that I had stored in a hard drive for years, but once I figure out the vocal harmonies, I intend on getting live singers to record them. So in my case, I'm using AI as a writing tool. To clarify though, the program doesn't create anything for me. All it can do is sing the words I enter using the melodies I create for it. My voice isn't great, so it's very discouraging when I try to sing them. Using the AI voice can also help me guide a potential singer on the vibe I want for a certain passage.
We're doomed. It was already bad, but now it's the fatal blow since we, musicians/composers, have become totally useless and the ultimate sinkhole for the average listener musical tastes. People endorsing this and making profit are the ones who should get deleted.
True it will flood the market and devalue all music. Also content creators are already using AI music/sound effects. Won’t be long until game studios, commercials, movies, shows and animations begin using AI music too
Ok, Doomer
This is sad
Pop music may be dead but hopefully most music fans have a bit more depth. Although the Depeche Mode one was pretty good and rather scary
But it makes you wonder what's behind the coding of the AI music, Like what happens when you listen to that in the headphones
All of this doesn't exist without human performers and to your point in the description laws need to be passed on this thievery - "Gramophone days" is obviously modeled on Frank Sinatra, not sure how you missed that - I'm sure his estate would be interested in contacting the AI company that "stole" his voice.
That's a good observation. All AI generated music is basically "photoshopping" fragments of known songs and pasting them together. It's cheap and grotesque.
That's what nobody is getting. All this "AI makes better stuff than us" needs to be rephrased to "AI is USING OUR STUFF against us".
@@heavenseek And the funny thing is that AI companies call using our stuff against us "fair use". This AI fraud will be remembered as the greatest scam in the history of computer technology.
@@heavenseek And the greatest mockery is that tech companies call using our stuff for free against us "fair use". I'm astonished.
@@federicoaschieri That's not how it works at all. The AI is trained on what "Rock" sounds like and what "Gospel" sounds like. Then when you ask for a "Gospel Rock" song, it starts with pink noise, and starts EQing the noise until it approximates what a combination of "Rock" and "Gospel" might sound like. The more specific the prompt, the more likely you will generate something that sounds like something else. This isn't meant to defend AI music, just to combats the idea that it is copy and pasting parts of other peoples music.
Anyone who thinks AI isn’t going to destroy music is in denial. I have used it and it creates better music than I can make myself. The singers are perfect, the bass is clean and hits hard. It’s already better than what humans can create.
It may be better than what some humans can do now, but right now it isn't better than most pro musicians. There will come a time when it is, but in this moment it's only better than newer musicians / artists.
I don't think it's going to cause much harm to real music. Many listeners do not want to knowingly listen to AI generated music. Fans want to see live music and build a connection with an artist.
AI artwork has been around for years now, and it's producing photorealistic images or artistic images in any style. Most professional visual artists are still professional visual artists. The AI artwork mostly only took away the bottom of the barrel work.
I don't think AI music will replace Human created music. But AI music does make it easier for people like me to create music without the hassle of looking for band mates to form a band. I'm good at vocals, but again I got no band mates to play with. But when it comes to lyric ideas and musical style to use with AI, I come up with the best songs that has came directly from my mind. And as long as the AI music comes up with original sound with the lyrics the creator came up with, only the creator should profit from it.
I just recently took my lyrics to my song and used Donna AI to create music to it. The result was a song so moving that anyone that listins to it loves it. Now the problem is that BMI at this time don't know how to handle it, they say their working on the problem, and that I'm not allowed to register my "partial AI created" song in my BMI Catalog. But Donna AI says that I have the right to use my Donna AI created song anywhere I want. I am going to use it as a demo to submit to artists for recording. I just don't know how that would work for a label unless the songs music evolves in the recording process to be different from the Donna AI version. But everyone tells me that the song is perfect as is as far as the music goes, so I'm not sure how that would effect the label and or artists interest in the song.
@@RobertHampton747 Yeah, I can see the roadblock in doing that. I have some friends who are musicians, but they don't play the style of music I used to create my songs. I'm thinking of going to use CDbaby or Distrokid to distribute my music albums with.
I prefer AI music. I’m done with human stuff
@@RobertHampton747 true. I myself want to distribute both my albums digitally. The thing is my music might be politically controversial. So I don't know of distributers will accept it.
@@philipandrews9211 yeah, AI has many upsides. I listen to my AI generated music more than the bands I normally listen to.
I love the analytical approach you have to things! Instead of "OMG AI will take away everything" you explain logically how we actually use generative tools already all the time. I personally think it's good for people who have no idea about how to make music and don't have the capacity to learn all that stuff (It's A LOT to learn)
But for me personally making music is putting my soul into it, and I cannot do that with AI, because I love singing and putting together my own puzzle pieces. Maybe I consider using AI for parts of the music, but making a whole song with it... Idk maybe it sounds good but it will never feel like "mine"
Thanks! I feel like there are a lot of sensational opinions on both sides, but I feel like reality is usually somewhere in the middle.
I can imagine a suite of generative AI tools that allow us to generate drum grooves that match guitar riffs, bass patterns to fit under chords, or allow us to generate 10 variations of a melody idea so we can more quickly find the best possible version of an idea. Or even just to kill writers block and give ourselves a starting point for a song idea. Or more sophisticated mixing and mastering tools.
Not everyone will use them, because some people like doing everything from scratch. But it will enable a massive amount of artists to make higher quality music than ever from their home, without needing to hire session musicians, buy beats or pay for studio time. They won't replace talent, but talented people will use these tools to become even more proficient and efficient.
@@AndrewSouthworth yeah exactly! I am really tired of having to scroll through dozens of HiHat loops etc, instead i would love to have an AI that can generate variations and choose the best, similar to what MidJourney does. I generate an image, it gives me 4 variations and I say which one I want or I want more like that.
Good examples that you said, so often I have a nice chord progression but I have 0 idea what kind of melody to play! Or how to make a nice Bassline to it.
Hope there will be something like that for Ableton soon, I would love that!
What is training data for a song writer to sit down and write a song. I will tell you. Every song they have ever listened to in their lives. Isn’t that unethical to go back on life’s experience to create a song, especially if they do not go back and compensate all of the song writers who had a copyright on the music. So tell me what is unethical.
The future will be people selling their NIL for others to create music with their voice. Its coming for the movie industry also, it's just a few years behind. The future will have people making their own Marvel, StarWars, StarTrek or whatever... Streaming platforms will have a "create" center where people can make their own Avengers movies using the NIL of whatever actors sign the agreement. If you didn't like the new Star Wars movies, well guess what you can make your own in the future.
Right now, it seems to me that yes, small artists will discover that they are drowning in a sea of sound-alike AI generated content. No matter what ingenious novel style they try to come up with. So, that's going to be a tragedy for the development of genuine musical talent and creativity.
But, also, it already seems to me that quality AI productions will be drowned out in a sea of middle of the road AI crap. Of course, that was always the big problem with the music industry. It turns out that it really is down to the terrible taste of most of the music downloading public.
There will be AI generated gems hidden away on RUclips with only 3 views.
And meanwhile I predict that the download charts will contain AI generated identikit Reggaeton and Afrobeat rip offs with the same rhythm, same basic bassline and horrible autotuned vocals.
It's what the people crave most, apparently.
First radio play of a Udio AI song was last night on Oystermouth Radio at just after 9:15pm. The show should be repeated tonight at the same time.
AI can make a good song, but AI can not make a great song
AI by itself won't beat a human, but AI + human will beat a human
After understanding this, i feel much better!
😂 stop thinking like a Boomer!!!! AI will surpass us. Just accept it. You won't though. You're stuck in the Denial phase...which I find hysterical. 😂
@@Recuper8 Thinking that AI alone will do a better piece of art than AI + Human, is being delusional, do your research!
@@Recuper8 You are obviously not an artist and know little about music. A good song is good because HUMANS enjoy it. There are infinitely more chances of getting a hit song generated by an AI led by a tasteful musician/producer than one led by an average Joe who knows nothing about music.
@@selliantuttimusi6735 Absolutely
@@Recuper8 AI is on track to hit a point where anyone can press the 'maek soong' button and make something 10x better than Mozart. But I think the most artistic creative humans will prompt it to make stuff 100x better. Our values will mold around this future but I am convinced creative people will continue to prosper in some form.
I think having some sort of identify verification system when you sign up for these generative platforms could be a solution. Possibly with watermarking on the creations as well... The same type of system you'd use for some distributors and banks... that way anything generated that's used commercially gets credited back to the prompter...and licensors can you use watermarking technology to disperse royalties.
I feel like it should be possible for them to encode an invisible audio watermark in the generated audio. Pick some frequencies in the 19kHz range and encode a binary fingerprint that nobody can hear, but you could see it in a spectral analysis.
Not sure how post-processing would affect this watermark, but it seems like a reasonable first line defense. I'm sure the people smart enough to code these AI tools can figure it out.
Sure but when a song is made it is pulling references from 100s of song so the royalty split between all those artists would be next to nothing
@@AndrewSouthworth There's a better solution than watermarking: force AI companies to upload to something like contentID anything which is generated. So everyone can check whether "your" music is really yours or AI generated.
@@ItWasntAPhase That's entirely a problem of AI companies. If they can't afford to pay artists decently, nobody will upload their music. I don't think AI music can be a business you can make money on. Soon there will be hundreds of AI music generators, since anyone will be able to build them, and they will all converge to the same result, as the technology is the same.
I'm kind of shocked these companies are charging so little for this. Maybe because they figure it will just be open source at some point and anyone will be able to do it, like Midjourney compared to Stable Diffusion?
I feel like if all of these services came out the gate at $100/mo for commercial rights it would be better for everyone.
1. It prices out a lot of bottom barrel spam creators
2. It gives more money to the ecosystem for a real royalty model to be created someday
3. The people that will actually use this commercially won't scoff at $100/mo at all, because they're likely already spending 100X that hiring humans to make music.
Let's say out of that $100/mo, $70 was allocated for artists who are in the training data. They multiply $70 times the number of paid accounts each month, divide it by the number of songs in the training data and pay it out on a per song basis.
Not sure how the economics of that play out but i'd like to know.
A very strange thing to me is that AI right now is very bad at hip hop. Like it will make hip hop but it sounds like it's from the 90s how the verses are structured and it will spit out corny lyrics. I think hip hop has an inherent advantage because it's more personal and based off unique experiences, at least not the generic kind. I know it will eventually close the gap but it makes me proud to be a fan of hip hop. For reference I've used ChatGPT, Claude, Suno and Udio.
I thought the same thing so I told my brother and he put me up on Red Roaring Lion on SoundCloud.. He is using Suno for Ai Hip Hop and I must say he really is trying to master it..Judge for yourself.. I followed and got on train early..lol
fuk Ai hiphop lol@@DaSpeciaList313Boiii
I already uploaded a full Udio album. I hate how good it is. I actually really like the album and I hate that I like it.
Udio owns it, correct?
@@AskDrLinqShorts not sure how it works. I don't even make money on my own music so not really worried about that. Just think it's crazy it can make music I actually like
I would like to listen to
@@marcosmontanhes AI Satan Awakens on my channel. It's mostly metal and hip hop.
@@AskDrLinqShortsAccording to Udio you can use the music you generate however you wish but you cannot sell it. Doesn’t say anything about monetizing it on YT or TT though that I know of. Many are doing that already
Artists that use AI will still need to understand what sounds good that the AI comes up with. Like if you write 100 songs, how do you know which one is the best? If your AI comes up with 1,000 songs, which one is the best? Or can you possibly release or use all of them? Big questions
No, no and no.
If you got a good ear you can tell the difference because there's no add libbing in the notes and no feeling because AI doesn't have any
I'm a creative person, and I for one have been having fun with Suno. I don't think what it produces is something that anyone would mistake for a polished studio production from your favorite band. And if you use it for any period of time, you start to notice a repetition of melodies and vocal performances. But for my purposes -- I see Morrissey from The Smiths walking around Manchester with a wild owl that starts attacking people -- I can go into Suno and write out lyrics and produce a song that makes me laugh. And living with chronic pain, I need as many laughs as I can get.
I wouldn't worry about how these models are trained, when AR Cloud matches and blocks anything too close to something else anyway. You would never know if your music is trained on one of these models if the output is up to par legally. Just as you'd never know if someone took inspiration from your music, so long as they made something original enough to literally be an original work in it's own right. People are getting way too upset over this aspect when there are literally a ton of measures in place legally over similarities in music.
When I first started making music 5 years ago, I started by covering songs in whatever chord progression and key I wanted. I used a modified melody. Then re wrote the lyrics entirely. By the time I tracked the song, and produced it in my style, with my lyrics, and my melody, in a new tempo and key, you'd never have any idea where it started. If you could, then the issue is with my interpretation of the song. It's too similar. The same applies for Ai music outputs.
If it too closely resembles a song, the output is fraudulent. Just as an "original" would be in such a case. If you don't want music inspired by and melodically and productionally pulled from other existing works, then you simply don't want new music. It's been this way before computers existed. Legally and creatively. We already have laws for the similarities of music. If you want laws over how music inspires and creates new music, then it should apply to real musicians too.
Never have I ever seen anyone ask for an upload checkbox to say "Allow other artists to use my track as inspiration for their track" [ ]
Doesn't matter how it was rendered, all .wav files are subject to these existing copyright laws, and these Ai gens already specify in their TOS you can't copyright claim other music based off your generations. They do not grant full master ownership for these very reasons.
Is there an option in these tools to plug in your entire music catalog and have it write one of "your" songs? Not that I would release it, but it would be a fun way to see insights on your own songwriting/production and understand the workings of the AI.
I've done this. It's not easy right now. You have to 1) have a lot of songs made 2) upload them all to the service whether it be chatgpt or claude, upload the lyrics, and audio if you can, then write your biography, your musical influence, your accomplishment, your goals etc. Anything you can think of to add more context to the music.
Then you can have it generate albums, songs, lyrics, ideas based on all the info you fed it.
It takes a long time, but it's fun and rewarding. You can use it to ideate or use it to see where you belong amongst current artists in the industry aka what type of music you make.
The tools are as useful as your creativity can take them. Good luck. Lmk how you did.
@@SixthSavioryou cant upload audio to ChatGpt. This is a weird response
@@Edbrad you can do it through the whisper API. You've never used chatgpt 4?
@@Edbrad You're a weird response.
I wouldn't do that without closely scrutinizing the terms and conditions of whatever ai you consider that with, as I recently looked at one that had small print that using their service meant you gave them non-exclusive terms to use your material to both train their ai, as well as use to promote their service, and so on, so effectively giving them permission to replicate you, learn you, and then give your vibe style to their subscribers, make new versions of your music etc without any compensation or ability to stop them, period. This is dangerous stuff, and artists need to step back, think, look over the terms, or they could be giving permission to be replicated, one they will never be able to undo once fed to any ai company and it gets out there into joe public's home computer and any bittorrent blockchain, etc. If it evolves to a standalone downloadable software not tied into any cloud, then that would potentially be only safeground to experiment with that, unless you don't mind getting replicated without any further compensation for producing 'new' versions of your music, and majorly without your even ever knowing of those future generations.
I am blown away by this song on Udio:
"Another Cigarette" by "MrTom12345" .
The song could compete with the best songs of all time in my opinion.
Good points! Imagine that a few years ago a singer-songwriter releases an album, but then gets visited by music business lawyers who want to see his/her spotify history to see by which artist the songwriter and his album was inspired and influenced by, and then force the songwriter give (a part of) the royalties to the artist he/she has been listening to. That would be crazy, right? I'm waiting for an A.I. Chris Lord-Alge app, that mixes my music exactly like CLA would do. (They should call the app Chris Lord A.I.lge, lol.) One year ago this would seem a futuristic fantasy, now I believe it something like that could happen within in 1 - 2 years. Just like A.I. session singers have appeared. (Audimee, is amazing.)
The mixing AI idea would be amazing. Izotope has been working on stuff like this for a while, but it's not fully there yet. I want a tool I can upload stems, and within a few minutes have a badass mix and master just as good as a pro.
It's also a lot less ethically questionable than the music generation, and it's a tool that enables indie artists to have access to world class mixing and mastering for dirt cheap.
I’m hyped to listen to your generative stuff! The plant music series is a favorite of mine.
I find it to be extremely useful to sample stuff, but yeah, weird times for sure!!
Damn. Better than I expected. One more reason that human musicians should emphasize originality and the human elements of their music. There is no training data for originality.
It'll be Original for about 10 seconds, because these parasitic models steal Everything.
AI shouldn't be used exclusively. There has been MORE THAN ENOUGH "song writing material" out here for inspiration. Artists used to talk about THEIR LIVES. The last fifteen years have been EVENTFUL as all get out. 2020 was like many decades in one year. SURELY SOMEONE can chose something to write bout from that?
choirs require a ton of money when it comes to recording for a track, do any of these platforms have a way to generate a choir?
Maybe but probably not specific enough for what you’d want. In the past I’ve used the Eric whitacre choir by spitfire audio for choir sounds.
People mostly see it as a bad thing if you think in a materialistic way. But once you look at it in a very complex way you'll realize it's got some bad but way more good
seems genre dependant anyone can program beats like this having mastery over an instrument is far more complex. It's like car autopilot we think self driving cars are coming but it gets exponentially more difficult.
AI Music cool.
Embarking on my musical journey in the 1970s, I found joy in creating and playing original music with friends, a passion that distinctly set me apart as I rarely ventured into playing covers. As the late 80s approached, my curiosity was piqued by the burgeoning scene of early techno music, a fascination that I explored through the medium of computers. This exploration continued into the 90s, a decade characterized by numerous jam sessions with friends. It was also during this time that I began to engage more publicly with my music, participating in open mic events and seizing opportunities to perform at various street events in town. Despite these ventures, my ultimate aspiration has always been to craft the perfect album.
My encounter with AI in music production has been transformative, offering an experience unparalleled to any other. It feels akin to entering a studio as a songwriter, where I am met with an extensive array of musicians and groups ready to bring my compositions to life. This process allows me to meticulously review each rendition, making decisions on whether they align with my vision for the project. This rigorous selection process is applied to all 16 songs, ensuring that each piece not only resonates with my artistic intent but also contributes to the cohesive identity of the album.
ruclips.net/video/3Gq7nrG9UPY/видео.html
Andrew I think it's time you think about what music you really love making, no matter the audience, and just make some real art. As good as generative AI is, in all of its forms, it has that zombie, mean-regression feeling. But a lot of intentional music does as well.
I just saw a famous but underground DJ the other week and it's like, the AI has no way, yet, to stumble to what makes his productions but also his live performance and ordering of music into a singular artistic statement. The true art will survive. You're a vocalist with a eurorack, f_ the algorithms lets hear Andrew Southworth in his glory.
I've been using it for a week.the stuff I've got it doing is wild.done a 30mins mix. Wow.ive got loads of years making music with my gear and it won't be long before I get bored of AI.
I like the opting-in idea where artists can get paid a percentage for useage. My publishers are currently trying different ways to work with these types of companies to monetize them some how, but it's all in suspense of what kind of outcome will happen for the future of artists. Hopefully, we get T Swift and Metallica to fight for equal rights for artists again if it really becomes a crisis.
Unless new laws are created, there’s nothing against listening to something and learning from it and then creating new music from it.
I agree with others saying indie artists will be hurt. There is going to be just to much 'noise' out there from this. The ability for persons to create their 'own' music instantly will tie up and clog the world with high-level engaging. People are linear and cannot do more than singular tasks and so I foresee there will be a small staple of industry hand-picked 'popular' human artists, while the rest will be replaced by ai. Already yt seeks to try to get labels to agree to a one-time payment for music catalogs in order to 'train' ai. What I see is this training will enable ai to replace real human artists with the goal of places like yt to no longer have to pay royalties. Without income, many artists cannot sustain themselves. But on a darker note, years later I see real human artists 'disappearing' yet their ai counterparts continuing on, and then a future where ai upped to lifelike hologram and a future world where people can pay/order their fav ai artists into their homes/lives with just monthly flat fee subscriptions as hologram versions, so effectively the music industry for humans is lifeless, just the way streaming has made it majorly unprofitable for majority of artists to even break even, this ai tech will go a step higher to being their own personal 'artists' or bringing their own favorite artists in ai form to them. Already I see people imitating artists in ai characters they are not supposed to do, but people break rules all the time, and this will bypass real artists and make it a joke to be a composer, a musician, when everybody can now be with just a few entries of words to ai already trained on high amounts of illegally gained content they are just not going to give up. It's going to be more and more a scene where we have to pay people to listen to our music, what a joke. To be a music artist now borders on being a vanity book publisher just from the poor streaming payouts and the way platforms are already inundated by tons of artists. A decade ago it wasn't that way. What a difference 10 years makes. Hate to think what 10 years forward is going to mean for new indie artists, such a term may even be past tense, as everyone may then consider themself an indie artist with the right ai subscription.
It still sounds trash. As artists, we see potential in what it generates and that’s why we’re spooked, but it still needs to be brought to life in the studio.
Quality with this technology is an S curve.. it will quickly assimilate to a certain quality, but that last 5% will be exponentially difficult for an AI to reach end-to-end. The vast majority of listeners listen to the top .01% of artists - so quality is everything. This technology certainly will add a lot of noise, but the spam will still be trash quality and ultimately lose.
Look at blogs, it’s very easy to distinguish AI content from real value content. When a consumer pays attention to what they’re actually consuming, it’s so obvious. And once you’re exposed for that, your reputation takes a massive hit.
Hopefully this content will get saturated to the point where people will get tired of it and call it out for what it is. Low effort content. If you are using AI to facilitate your art... That's cool... If you are using it as a tool to enhance the time, effort, and creativity that you are putting into your creation, then that's fine... But if you're simply going to a website and entering in some prompts, and then spitting out a song that was generated by ai, you shouldn't expect to be monetized. These types of channels should be demonetized. RUclips is about to require content creators to state content was "synthetically generated" but how will they enforce that? I have already come across these comedy parody RUclips channels where they are disingenuous and straight up lie to their subscribers. They literally say that these songs were created by musicians. Or that they found some old record... And people that are not as informed are believing this. Something needs to happen.
I've written my own lyrics and used AI to generate the music and vocals. It can be extremely frustrating because I'm, currently, at the mercy of the AI for the sound and the melodies. But, I finally, after 2 hours, was able to hear a cut that was very close to my idea. Unfortunately, after all that I wanted to change a couple of lyrics, add new verses, but I couldn't. I'm sure those sort of changes will be available at some point, but it can be useful for someone like me, who's had ideas, but can't play the music or afford to pay someone to create it.
"Something needs to happen." Yeah, it's called "universal basic income".
@KhanumBallZ I think you meant to say all AI Art?
@@rdgtxs yes, your circumstances unique. You're putting a lot of time and effort into it, personally.
@@Recuper8 well, I don't disagree with that.
I want to feed my demo to AI telling to keep certain things (lyrics, melody, main harmony etc) and prompt to alter style and sounds.... kind of fast producing.
Could focus more on composing and let AI to produce different genres....and with different singers (Elvis, Chester, etc)
As owner of the song I would be safe and possible to reach new audiences.
Graphic designers and producers....basically all those making the endresult shine are propably the loosers here. Sad but true.
How long do I need to wait?
6 months is my prediction.
Ps. Still learning to use all plug-ins and stuff despite knowing soon to be almost useless knowledge.
But hey....try to do very specific image with Midjourney. Impossible. Same will happen with audio generating so keep you producer skills tuned still😊
And thanx Andrew for this great channel. Makes world a fraction better.
Big companies dont need to pay artist anymore , thats gone happend
I'd think the better this gets, the less opportunity/value of real artists. However, I think it could still be used as a tool. It will just become a competitiom of who can manipulate AI best to get the best song out of it.
I'm surprised by how good it is and have a lot of fun, however I still value human-made music over this. Nothing beats natural-creativity that's expressed through music.
Another important point, is about music composition. I can understand if AI become insane producers in term of quality etc. But when it comes to melodies, there is no mathematic formula to make a great melody with emotions, it's all about feeling, and AI don't have that. So there is also hope here, i hope those tools could become much more customizable and tweakable.
I don’t know about that in re: AI melodies not having soul. I mean, yes, I guess you could argue it’s not literal soul, but that 1940s style female vocal track Andrew plays in this video sounded like it had soul to me. If you walked into a cocktail bar and heard that playing, I don’t think you would think it wasn’t a real person with a real soul who had made it. Another argument is that AI music is an amalgamation of many pre-existing souls combined- real people who felt real things and made real music.
Hey Vanessa, thank you for your reply, i agree with you that this example is pretty good! Personally i would say that the vocal melody is ok, not super great but it's indeed impressive! BUT, to me, it's the voice that is making it much more emotional, it's really impressive, but the thing is this voice come from someone else voice, it's not a 100% synthetic artificial voice, it come from human data, and i am pretty sure that the first problems that will come related to AI music will come from the voices, artists will hear their voice in some AI songs, it will be a lot of sues i believe, (there is already many examples, and it's already happening). So, yes, there is emotions here i totally agree with you, but it's like cheating, cause it come from a real person voice.
But if you do the test, you play just on piano the SunoAI generation, you translate them just with piano, you'll hear that many often there is no senses in the melody, it's not what a human would have done (usually).
BUT, i believe that through a ton of quantity generation, you can with LUCK (and no skills) have a beautiful composition
The thing is, it knows what melodies humans find emotional or powerful, so it can generate new melodies that also feel the same. You can genuinely make tracks that feel like they have so many feelings behind them, pain, love, happiness etc - except it's just an algorithm engineered to make you feel those things.
It seems like at the moment it's actually better at capturing the melodies than the production quality.
Apparently on Udio's roadmap they're planning on allowing people to download the stems for their tracks, which is a huge leg up in making these sound great or using them in more useful settings. Apparently you'll also be able to use it to generate samples and sound effects, which is also super useful for a myriad of reasons.
Yeah from my tests, the average person can't tell the difference between human made and AI music. We might be able to tell, but we're all actual music creators. We buy fancy speakers to mix our music when 90% of consumers listen on crappy headphones or their phone speakers.
I don't think AI music will ever be better than the best humans in the world, but its already making better music than many humans in the world. I think humans will start finding new avenues of music that AI can't do well and dive into those areas. Basically cookie-cutter music will now be mass producible with AI so humans will be forced to get more creative and innovative to stay relevant.
I think human artists will also have to be even more open about their process, and share even more of their personality. Because the process and their personality will be what makes them shine and stand out for the sea of AI music. In some ways I feel it could create a renaissance for music.
My music is all AI done.... and I'm trying to become one of the first wave of AI assisted music artists to reach a 6 figure income from their music. So I'm doing the climb one step at a time with the music to try and get there.
I don't understand people against AI. As an artist the important thing is making use of what we have to make new art. I'll leave all the questions of behind the scenes copyright law to lawyers. But right now everything 100% legal and legit.
Don't call yourself an artist, you understand because you aren't a real musican.
You made a great point. The only way AI music can be legal is to tie it to distributors, and AI companies will have to offer yet other platforms that we upload music to, if we want. That's the only legal solution. That's what will happen, as it's not true that AI companies do what humans do: in order for a human to be "inspired", they have first to consume legally the music, so pay the copyright holder for a license. AI companies have to pay as well.
That being said, I don't think it would be smart for labels like Universal to license their music to tech companies, unless they have a partnership with them. Suno is just 12 people, so literally any label can make its own AI and earn money exclusively. Once a technology reaches mainstream, it costs nothing to replicate. That's the self-defeating nature of AI: since the machine does all the learning, the human factor counts nothing. So again these are all the premises for a new AISpotify, with which labels will be partnering.
I don't think Suno et al. drastically dilutes the music pool anytime soon - you still need to take these ideas into a DAW and re-record and re-produce/program everything. In other words, you still need to know what you are doing to get it to the right level of quality.
I also do not see "average" people getting heavily into Suno anytime soon either - even if the quality is there, you still need to hop on the app and enter prompts. It requires work. Humans are wired for efficiency/laziness and going on Spotify and having a playlist made for you is still easier than that.
The better use case for Suno right now is as a writing tool for existing musicians/producers.
In a year or more though? Who the hell knows.
It is also being used by content creators for intro/outro/background music. I can see game companies use it, smaller tv productions and maybe even ad agencies to come up with ideas
@@ItWasntAPhase definitely
After Hip Hop/ Rap what did you expect?
Honestly, outside of the scale and speed of it, how is A.I. learning from music any different than humans doing the same thing? I've I've never played Jazz but heard it and liked it, then I went out and listened to every Jazz song I could find and allowed that to teach me how to write a Jazz song, would that not be the same thing? Not trying to be sarcastic, just I honestly don't see the difference.
In defense of AI inspired music I would like to remind everyone that every piece of music that has ever been created was inspired by existing music. When it comes to AI generated music I have to assume these completed pieces are being inspired originally by a few lines of text, followed by a multitude of bits and bytes the computer puts together to create a song. So leave it alone because what we as humans will end up with is a better musical product that still has a human element attached to it.
Every time AI shows up in a new space...the people in that space make the same mistake. They conflate the "end product" with the "AI model". They are act as if they are the same. They are not.
The AI models are fair use. Why are they fair use. Because they are transformative. That is...they don't resemble the original in any way...they don't have the same use...and they don't compete in the same market. It's transformative.
Transformative use is fair use and fair use....well...you don't HAVE to "ask permission" to use someone else's work.
The math models don't resemble the original works in any way. So they are fair use. Now one could say, "Yeah but the people USING these software are "ripping off" the original artists."
Well...that would be a case that would have to be made against every single person USING the software. You'd have to be able to convince a jury that a udio user had made a "copy" of your song. But you can't go after udio itself because they made a tool that someone MIGHT use to copy your song.
That would be like trying to sue Adobe because they make Photoshop and bad actors use Photoshop to violate copyright all the time. In countless ways. Until you can come up with a reasonable line of thought on why Adobe should be able to be sue because of the ways in which "some" people use their product...you cannot make a reasonable case that udio or OpenAI or any of the rest CAN.
(And please don't start with the "but they're making money with it so it's not fair use" misinformation. There is nothing in the fair use doctrine that says you cannot make money when you use material in a fair use way. Otherwise all those satire cartoonists in The Atlantic that poke fun of major copyright IPs like Star Wars and Marvel would never be able to be paid for their art. But we know they ARE paid. Even though their art falls into the fair use category. Why? Because you can get paid for your fair use work.)
The piece that I argue should be controlled is not the usage of the output, but the data used to train the model. Everyone should have the ability to choose whether they want their IP used in a training model that can be used to output new creations.
Granted, you can't choose if someone makes satire of your work. But these generative AI systems are a new beast that I don't think should have the same 'rights' as a human.
I feel there has to be a legal discussion between all the various stakeholders in the industry behind what rights and safeguards we want these tools to have. At the moment it's the wild west and it will take governments years to decide anything, but that discussion needs to happen to decide what is best for everyone involved.
Maybe the answer is as simple as all these tools must watermark the output in someway so that it's easy for everyone to tell if the audio was AI generated. Then platforms display all AI work as Ai generated, and the consumer can decide for themselves if they want to support or skip listening to AI generated music. Perhaps AI generated works get paid less due to the ease of creation, and it prevents royalties being taken away from actual human artists.
Or possibly the answer is more complex, and we decide all training must be on an opt-in basis only. And people can negotiate how much they want to be compensated for having their work be used in an AI model.
To be honest I think ai music sounds as artificial as AI images. I made a lot of songs that sound great but after a while I started to get nauseous and my ears hurt after listening to them. I don’t know which frequencies they have but they male me sick
AI-generated music has made impressive strides over the past six months (since this video was published), but it still won't replace real musicians. Live music offers a unique, irreplaceable experience that goes beyond just the sound-it's about connection, energy, and atmosphere. That said, for jingle writers and stock music producers, the story is different. AI has essentially made their roles obsolete. But let's face it: nobody's making real money from streaming anyway, whether the music is created by humans or AI.
Imagine you subscribe to a service like the one you are describing. How can you prove that your song(s) has been used to create another song and therefore you deserve to be paid?
To me, these AI-generated songs sound soul-less, and yes like they've been done before. I'm curious if anyone has analyzed the sound quality, loudness, width - essentially the release-readiness of this crap?
The songs I've generated I found I still have to master. They're not release ready. I've used Suno and Udio.
I ran what I created through an AI mastering tool and the results blew me away
I've gotten it to generate really good sounding math rock songs that are way more moody and even have fascinating song structures. It's horrifyingly good. To clarify, I'm not using it in my work at all - it was just an exploration.
@@nova_harbor good to hear it was just an exploration 🖤I believe AI could potentially be a good thing for somewhat more "creatively daring" musicians (e.g those artists not trying to sound like someone else) which could see a lift in more experimental and niche music. That would be cool.
having spoken to these AI "filters", no, they make zero decisions, they type it in, grin because monkey grin at funny machine, then dance when the music comes on and say they are "producers", so, they are less than knob twisters.
They're definitely not 'producers' and most of them are just acting like monkeys in front of a computer. However there are also people who spend a lot of time perfecting their prompts, generating dozens of versions to find the perfect one, using other software tools to convert them to stems and then mixing / mastering the stems while adding new layers to improve the track.
Just like actual artists, there are different levels of work people put in. Some artists buy a beat online, lazily freestyle rap over it and then throw it online. But other artists make everything from scratch, obsess over the writing for weeks and properly mix and master the song.
Ai music, a lot like Ai in other industries, will not kill music creation. People will always want music from people they can connect with. Problem is AI music will flood the market and devalue artists. Also the landscape of music in commercial uses like games, commercials and content will drastically change
The like the idea of the corporate entity having to pay a subscription to the artist for training data. >:)
I don't think there's anything wrong with it. A lot of artists try to sound exactly like other artists. It's called inspiration. It's not stealing samples from anyone. Samples and presets are different things.
Okay i am way past the "Fuck, i am angry." I more into, how do i make this work? I mean what do you as an creator use this and give your spice to it? It's got all the instruments and songs. Do you sing with it or what?
Great idea but keep in mind that the tos says you can use the music however you wish but you are not allowed to sell it. So the songs you create can be monetized to stream but now sold as tracks or an album
Don't give in and join.
yeah great idea.. let's all sing with it.. bright future ahead of us everyone! nah fk it, im not laycing my music with it, who knows what kinda subliminal messages is the algorithm putting in 😬🤔
You talk about companies having a "right to train" their A.I. models on pieces of music. When someone goes to school for music, are they only allowed to utilize music that the school has acquired some sort of approval to train their students on? Is a teacher not allowed to make reference to a song in order to teach music theory or composition if they don't have approval from a record label? What is a "right to train"?
Exactly. "Right to Train" is BS.
AI cannot be 'inspired' by something, because it is not capable of independent thought or feelings. It's not taking inspiration from songs to create something completely new. What it's creating is an amalgamation of it's reference data. So in that case, it could be argued that musicians DO deserve royalties for their music being used in training data. If you're building a house, unless you make everything yourself, you have to go out and pay for the bricks, the glass, etc. You don't just get given the material for free.
I just made a collab with AI. It's a video up on RUclips now. I wrote part of the lyrics, then the AI wrote the middle stanzas.
The song I did before that was done with a plugin where the rave instruments (that I don't have) made an entire song for me if I just pushed the buttons in a row every 8 beats. What both were doing, were playing popular note progressions, ones that historically sound good, and that we equate with good music, what we want to listen to, with great tension & tonic motifs. These riffs are in a lot of techno songs.
At first, I felt like I was cheating; I'm just a guitar player with some minor piano skills. I love the latest 2 songs I did, but I didn't do them; it is weird.
When I saw the ad for 'Donna' I got excited because I thought I could use it to make references of songs I've written myself. Finding qualified reference singers has always been a thorn in my side because I've either run into singers who only want to work on their own music and are unwilling to help out - even when paid, or I haven't had the amount they ask at the right time. I saw the A.I. app and thought it was the answer. But, all it does is create its own songs. Yuck.
Songwriting is my first love. I take pride in stringing words together and creating songs with my own mind. It's like instant grits. We all know how My Cousin Vinny and any self-respecting southerner felt about instant grits. It's just not quite as good. Where's the artistry in an app like this? Where's the soul? This is nothing more than a dream killer. But, as stated in the video, it's here to stay no matter what we do.🤬
As a venue for music I would never book AI music acts. Just out of ethics. If I get the slightest hint that they are just AI based - off you go. Maybe won't take long until there are careers built out of this AI covering of songs but I guess we already have a lot of repetition of the samey old shit.
The worst part of this is. It devaluates human creation with its natural flaws and beauty. Cause what these songs need to do is to appeal to people who just want to throw out these working songs. If they would be just amateur or something like that people wouldn't use this. So there is an inherent aesthetic and hearing habit that was put into there as a bias.
Be nice and support your local human creators.
The Artist have to go the modern way right now! Write own Songs, create AI Songs. Get Ideas from AI and Remix ur own generated Song.
Of course its harder to get into the game now but the listeners are blessed with new nice Music. I think its a win win on both sides!
I think AI Music is really good for those who feel they dont have what it takes to live there dream, im a singer and a lyrisist, i have no band and i intend to sing on my track that i make, i just dont have a band, im nearly 40 so finding bands im into in my town at my age is hard, NU-metal with pop vibes
Not true at all.
I feel like this is going to ruin singing artist or producer and here comes the prompt artist replacing producers all kinds of stuff or maybe we just have to raise are prices and put value in our music and lower prices in ai
100% I'm looking to save money by getting AI to sing hooks for me instead of getting a feature artist. I do everything in my production process except I can't sing. I make hip hop music so every now and then I'd need a singer.
There is a logical basis to what you say about how the artificial intelligence was trained, but the only thing that ultimately differentiates the artificial intelligence as a music producer and the human as a producer is that the human chose and listened to 10-100-10000 songs from which it (possibly) drew inspiration (imagination - influence) and the AI was FEEDED (so it also listened, but without conscious participation) 10-100-10000 songs from which it drew inspiration (imagination - influence) and created something new. I will take you to the case of theft of intellectual property of the invention of car wipers, where the inventor, when accused of having used EXISTING electrical devices in his invention (so "it should not have been accepted as HIS invention", but as a different use of existing inventions that gave the effect it gave), he answered that the words used by poets in their poems are also given, but this does not make one poem a product of theft from another on the grounds that the same words have been used in one and the other. Likewise, here too, we all know that when it comes to music, there is no parthenogenesis. We have all been influenced by something (either consciously or subconsciously). I find nothing wrong with the exact same thing happening with artificial intelligence, even if in its case, the influence comes through human intervention.
Generative AI models are trained on copyright-protected data - is that legal, is it sampling, scanning, theft. How can i protect my art. How can you say no to your art be scanned by ai???
You cannot copyright music made by AI.
That might be true, but you can still commercially exploit an AI track. Also as far as I know that's only for registering copyright, which most artists don't actually do anyways.
Hey Andrew thanks for the comment. I’ve enjoyed your content for along time. Yes, that is true. Also as part of the Udio agreement which you sign when you register you have to give them credit if you’ve used their platform to generate a song.
Over 6 million songs have been generated by Udio in the last week. If those songs are uploaded to streaming services it further dilutes the streaming payments to artist that create without AI.
Yeah for Udio you have to give credit, not sure if their new pro plans remove that requirement. With Suno you do not have to give credit and as far as they're concerned, if you generate the audio you own it fully with no strings attached.
We have a small window of opportunity left. AI isnt that soohisticated yet so I say maybe 1 year? LOL 😅
😅
Not even...
This is scary but I've tried it. I play guitar and keyboard and sing. I'm gonna use this as a tool. It helps me with ideas. If my AI songs become hits I will make sure they know I used AI, I've used chatGPT to create lyrics but I change them more to my style.
oh brother..
Those old timey ones are lit!?!? ❤😢
But you see how little time it took him to type in those words to have the AI generate that? It isn't right for people to be monetized for putting in no effort meanwhile RUclips content creators are putting in tons of time, effort and creativity.
@@ChrisCeeKayKelley It's not even an act of creation.
My own organic intelligence was trained on copyrighted material. As was everyone else's.
You're not learning how to do something and you're not creating anything. You're placing an order with a copy machine that imitates the creations of others to manufacture you a product.
The bass is slamming, too . AI pays attention to the BASS
Composers and music producers need to get the courts to force AI companies to give each of their artificial creations a registration number. This number must be saved permanently so that the AI song remains identifiable as such forever. And therefore NO copyrights can be claimed!
You talk about companies having a "right to train" their A.I. models on pieces of music? When someone goes to school for music, are they only allowed to utilize music that the school has acquired some sort of approval to train their students on? Is a teacher not allowed to make reference to a song they in order to teach music theory or composition if they don't have approval from a record label? What is a "right to train"?
Treating the AI like a human makes sense in certain scenarios, but they should absolutely have to get permission for training. No question about it.
Huge difference between a music student and a machine that can crank out thousands of songs per day. This isn’t a sentient creature we’re talking about, it’s an algorithm made for profit.
@@AndrewSouthworth So what if it's not a sentient creature? It simply does what a human can do on a super human level.
And even if it is made for profit, that's what a human does. A human looking to enter the music industry trains itself and then produces music for profit. Often times these small-time music producers don't hardly know any music theory. They just know how to create a four-chords song that sounds like a popular artist, then get revenue from it because the masses have been trained to enjoy repeated four-chords songs.
Or what about music libraries. Is making simpleton, stock music for advertising really that much of a dream job?
Suno and Udio are not stealing the music any more than a human is when it's analyzing a chord structure or riff or instrumentation by ear. They're not distributing it or claiming to own it.
All that can really be said about these companies is "it's not fair". Personally, I think it's already a crime that the music industry has made as much money as it has off of such formulaic drivel.
That's my opinion at least.
In my opinion, if the company is going to release a commercial product they should have to get permission from the artists to use their music for training purposes.
When a human learns music and get's inspired by artists, they actually pay for music. They stream it, go to shows, buy tab books, take lessons etc.
The AI does not compensate the artists like a real human would while learning to make their own music. At the very least, they should be required to at least purchase every song in their training data as a digital download.
@@AndrewSouthworth Okay I can see that. They pay for the music, like as a digital download. And then maybe they sell training data perhaps bundled by genre. And any music a person generates from their purchased training data they have full copyright for.
As of right now, Udio is not selling their A.I. service. And I don't think users have copyright for the music they generate. So maybe if it stays this way it remains a free service, but giving the users copyright over generated music would require the purchasing of training data.
That makes sense to me.
Training is not how you think, also is possible training it self. Even training on artists music is the same as if someone just heard the music Notting more
I actually believe AI is an advantage to real musicians because they can use AI to be inspired and then repeat the song using real instruments
100% agree, like Pinterest for chefs
Are we forgetting that the usual case is one human makes better songs than another human?
I can see this as a threat to those who produce cookie cutter pop and rock, but do you ever see it get to the point where it can make a genuine issue Stairway to Heaven, Paradise City, Bohemian Rhapsody or even My Dark Twisted Fantasy? Certainly don't foresee it able to create a Brandenburg Concerto or Pastoral Symphony. Generic music yes, monumental works of musical art, not seeing it yet.
The drum sounds alone are of the same quality as in the 1970s. Guitars are difficult to track and are of poor quality. You can currently only use this stuff as a suggestion and replay it. But then you're very poor if you can't think of a better song.
its the beginning of the end mate/// predicted it last year already. Music industry will be full destroyed by it
I don't think so. People thought recorded music would destroy the music industry, then radio, then physical media, then synthesizers, then autotune, then digital downloads, then streaming etc. The people that care find a way to keep doing what they want to do, and the business folk keep finding ways to make money doing it.
@@AndrewSouthworth yea but AI is something else mate. Radiomusic and all other music came from humans and now a robot can do such melodies. So bye be music industry as we know it because any human can make kinda good sounding music with one mouseclick. time to focus only on live performances with instrumental and vocals.
@@rolandvonutopia It is making music trained on other music. Human music and creativity had to exist for the AI to function. If you make boring cookie cutter music, yeah, AI will probably replace you. If you make interesting and creative music, AI can't touch you. AI is not creative.
Yes this is sad
I would say - don't let the companies scrape all the songs. As a legislator I would get into their data system if I have authority and check whether they used copyrighted data. Which they definitely did because the open songs that are on the net are not sufficient enough to create such vast amount of styles and proficiency.
Wonder who will win in the case. Same as the visual art. I have the suspicion that with legislation there will be a lot of court cases.
I am against the creation of these songs as a legitimate art form. They are not art. If you as a human are not putting in more than 50% of your own creativity then this is not art.
I have Donna AI Music/Lyrics Generator and I Love It and it's Addicting Plus Everyone Loves the Songs I Create With My Ideas 😁🥰