Then don't listen to pop stations. There's thousands of hungry solo artists online, trying to sell their original work and go without a recording contract, but most people can't be bothered to spend an hour or so searching for good music online, and instead whine about what they're missing, when a little effort will find you exactly what you want. I'm in my 50s myself, and I can't put all the blame on younger generations--technology has made people my age lazy as hell, too.
@@rikk319 yes. Agreed. We are getting so lazy we don’t even think anymore. Good examples are the karens and kens so full of entitlement and they don’t think before they become YT superstars by opening their mouths without thinking.
Maybe one day AI could be used to create a band and music with members that YOU choose. For example, Jim Morrison on vocals, Steve Vai on guitar, Chris Squire on Bass and Bill Ward on drums, and then it creates a song using those band members. That would be cool ( and kind of weird lol ).
I think there was a cheesy Righteous Brothers song about that in 1974: "If there's a rock 'n' roll heaven, well you know they've got a hell of a band...": Jimi gave us rainbows And Janis took a piece of our hearts And Otis brought us all to the dock of a bay
For once I would love to see that happen, but I think this is a hell we have entered without our knowledge and no matter how hard we try to check out at any given time, none of us, not even Don Henley can never leave.
Fil, I am old enough to remember the voices of The HAL 9000 and Max Headroom, so to hear a computer sing a song that it wrote with an American accent, blows me AWAY!
As you mention … Thing I’ve found interesting is modern pop music sounds so formulaic, quantized, tuned & processed that makes AI music sound so natural , and sometimes even more enjoyable than current pop hits
@@morbidmanmusic You can keep your fake "...ride... ." If you don't care that a voice has been pitch corrected, or that a piece of music has been auto tuned, that's your prerogative. I prefer my "...ride..." to be an authentic, 100% real one - every time.
I@@morbidmanmusicI think it does matter, as music is fundamentally a form of human expression. It is an art. I know modern pop is more product development than art, but I still don't like it. There are things AI can usefully help us with in scientific and medical fields, for example, but it is not needed in the arts.
Both songs sound surprisingly real. If something like that was played in the background in a shopping center I would never have guessed it was AI generated.
@hws3044 Semi-sucked, musically, maybe but it was a hit of sorts, and unforgettable...hence your remembrance of it at all. Portending the future that way was jolting but closer to the truth than we care to admit. It was a statement on the downward slide we've been on since the industrial age. There should be an update of that song because humanity, at its current pace, probably won't make it as far into the future as Zager & Evans sang about.
@garytwitchett9359 i really dont understand how anyone can think that song sucks, because it is deeply moving worrying just slightly overwrought. this just sounds like millennial verging on genz beavis and butthead. next year the song will be cool because king gizzard and the lizzard wizzerd do a cover. these people are so " ooh look at me being edgy, oh but now someone i appreciates tells me it is good". have this all the time at the radio station. anyway the song isnt anything like the year 2525. it s a shitty contemporary rockist song. it is at the level of nickelback cringe. and yes very technically accomplished band, none of it moves me. so the whole argument starts from capital B capital S . (not very) respectfully .
@cat-star5403 lots of musicians start out crappy and then get better. It's an opportunity to have fun and encourage new talent and maybe have a story to tell how you knew them before they hit it big
Easy enough to say that, when you know it's AI. Fair enough, you can say it doesn't blow you away, but there's plenty of human made music which doesn't do that either. And this is one of the more bog standard efforts I've heard from AI. I have heard some incredible results elseware. ... Not saying I like it. In fact, it causes me great concern, but it alsoimpresses the hell out of me. And some of the songs I've heard do sound fantastic.
“No offense to the AI generated voice” 😂 Fil, get ready for AI troll corner! Man, those lyrics too! 🤣 Good point about the AI sounding more human to us since humans sound more mechanical these days.
Pretty sure they are generated. At least with Udio that is the default, tho you can supply lyrics. Generating lyrics, poetry, etc is something AI can do. Likely not great lyrics... But lots of people are having fun using these things to generate music about particular odd topics, memes, etc
At 79 I've learned not to fear technology, just enjoy and use it. I go back to the first studio I worked in, in the mid 60's where we had two 2 track Ampex tape machines pulling tape at 15 ips and an 8 channel tube board. No punch ins. You had the whole band playing each take and a "punch in" was taking two or three takes and using a splicing block to pick out the best parts of each take and cutting it together. And you tried to minimize those cuts. A good band could do a seamless take with no cuts. And then an overdub would be to run the first pass thru the board, add the vocals or whatever else you wanted to "overdub" and catch it on the second machine. You could do this any number of times if you needed to overdub several parts, but you lost a bit of fidelity and the tape noise would build up with each pass. But those Ampex machines were really good at not degrading it too badly. But you still didn't want to go past 3 or 4 generations. The most serious flaw working like that is you could not go back and remix it. After playing in bands for about 20 years, the tech started to give me some tools to work with and I became a one man band early on using a drum machine, left hand bass on synth and right hand keyboards using a DX7. Almost no one was doing it at the time and I made a very good living replacing a band. Then I opened my studio and went thru the 388 Tascom 8 track, then Alesis 8 track vhs syncing 3 together for 24 tracks, then the hard drive Alesis 24 track, and to the present with a very fast computer running Sonar (It's now Cakewalk but I was using it long before it was free) into 2 Hammerfall 24 track interfaces into a Yamaha DM2000 automated board for 48 tracks. Of course you can use buses and submixes in the software if you need more than 48, but I never have. I think the most tracks I've used on one song might have been in the low forties. It works like a charm and a dream set up compared to those first Ampex machines. The Ampex machines had a sound like no other, but the limitations were far to great to overcome like what we can do today. I now supplement my retirement making videos for youtube and I still do a one man show for my youtube vids unlike anything I've seen on youtube yet. And yes, I do play all those instruments. Not at one time like I can on the vids of course lol! So you ride the wave of tech, using what it gives and making the best of it. But it still comes down to what you do with it. I'm amazed at what I hear with the AI, but I'd still only use it as a tool, much like a drum machine, to get ideas but I'd still redo it and add the human element. I would definitely do my own vocals, and I don't use pitch correction, make whatever changes that seemed fitting, and I'd also replay the parts and do my own mix. So just like with the various plugins we have today, it's how you use them that makes for individuality. I guess what it's coming down to is who will be the best at using keywords and how creative a description you can use when asking for a track. I don't know how long it takes the computer to give you the "finished" product, but I can see how you could have several ideas to play off of and then come up with something "original" lol! I am fascinated by all this, and will have to look into it myself. I'm an old dog, but I'm always on the lookout for a new trick!
Thank you for that bit of history. It's fascinating to look behind the curtain and see how the music I love was made. I'm an old electronics nerd myself. When I started the world was using vacuum tubes. I've always had to constantly train (as you can imagine). At the beginning of my career, I could repair almost anything out there. Today, the technology is advancing so quickly that it's simply impossible for anyone to keep up with everything. I enjoy watching this explosion of new ideas. Technology is always going to advance. You may as well enjoy the ride.
The vocals are very one-dimensional AT PRESENT but in a matter of a few nano-seconds it will have become so much better. Take me back to 79 this world ain't for me.
There are bands that are playing "modern" 79 music. Check out Elephant Stone. Such an amazing throwback. The latest album was released a month or two ago.
You can't go home again. Instead of dreaming of what can't be, vote with your pocketbook and stop supporting megacorporations that put profits over people, and buy at small mom and pop places whenever possible.
When people are shocked that a group can be offered a billion dollars for their music catalogue (looking at you Queen) it isn't entirely the existing songs that they're after, it's the right to generate new music particularly for movies and advertising using AI. Interesting times ahead for us.
Nightmarish times. Nothing being pursued now has anything to do with art, or even improving the world. All of this is just in the service of wealth, and it isn't going to just be art, it will eliminate the need for every position it possibly can.
@@EdwardRLyons They made a truly impressive legacy for themselves, but if it hadn't been them it would have been someone else. The push to liberate finance, and maintain American hegemony was already well underway.
@@EdwardRLyons terrible take. Probably the exact opposite would be closer. This is about what big corporations will do by replacing creative people with bots.
William Gibson (author of Neuromancer) predicted all of this in his 1995 novel "Idoru" (Japanese pronunciation of Idol). One of the main characters is Rei Toei, an AI rock star. She has been programmed to remind viewers of their favorite J-pop idols. Implicit in her design is that she is not one Idoru, but many. Individual viewers and fans will have a personalized Rei Toei album, video, and collection of images, as 'she' can be and is customized according to the tastes of viewers.
@@ettinakitten5047 Conceptually similar, but Rei is much more sophisticated. Rei is an actual AI which customizes itself for each person, and generates music in the genres they favor.
I've read the book, and I placed it firmly in the sci-fi realm in the mid 90's. Now, 30 years on, it's frighteningly close to becoming a reality. Maybe it's time to dig it out of storage and give it another read.
@@thePrisoner1000 I do own it, but it's one of those books that I keep taking off the shelf, and putting back, saving it for another time. Maybe it's time to read that one as well.
people always say milli vanilli . kind of sad. because the whole of boney m by the same producer were equally studio musicians. they tried to get more credit and failed in favour of the actors. the death of ( ithink ) rob (vanilli) made it even more tragic. didnt stop frank farian.
The lyrics are hilarious on the second one. I don't know that I would have picked it out as AI, because most things recorded in the 21st century sound like machine generated drivel to my Luddite ears, even ones meant to sound like something from an earlier era. The copyright question is an interesting one, one you could also apply to AI generated art and books; at the moment it seems to be a bit of a free-for-all.
I asked Udio to make me a prog rock instrumental and it's rather impressive. Udio has you stitch 30-second sections together made of various arrangements that are designed to flow into the other saved sections. This is fascinating and bewildering technology.
Udio is good, but does suffer from AI coherence issues, quality much better than Suno, but big questions over ownership with anything generated in Udio, where as Suno, is clear, if you pay, you own the rights, already talk of Udio breaching copyright, in the way they have created models, be interested to see how that goes.
We’ve only had recorded music for the last 150 years, and modern popular music is already so heavily produced that it might as well be AI. Though I think AI might make it even more difficult for musicians to monetize recorded music, I think live performances, where the musicians and the audience react to each other, are magical and will hopefully still be a way for people to experience music and for artists to make a living.
I love new music, and there are brilliant new artists around who play and sing authentically. You don't have to be a grumpy old git to dislike this erosion of art on music.
@hws3044 Unless you live in a cabin in the woods with no running water, no gas or electricity, grow your own food (from scratch, without buying any seeds), make everything you use from scratch, and never use money to buy or sell anything, then you're going to be using computers in some way, or at least the resources and items you use will have used computers in their production. You can refuse to use a PC, not own a smartphone, and pay for everything with cash, but that doesn't mean you can escape the influence of computers.
@@GeeEee75 Another reason I hate them. We did everything before you can do now, only faster, with human interaction, and we didn't spend all day on the phone with some tech guy, trying to make our tools work so we could get a job done.
I'm reminded of songwriter singer Neil Sedaka who had some early success with pop and novelty songs but feared he was heading to be a one-hit wonder. Sedaka then bought the three biggest hit singles of the time and listened to them repeatedly, studying the song structure, chord progressions, lyrics and harmonies before writing his next songs. His next song was 'Oh! Carol', followed by a string of 50s, 60s, and 70s classics written from himself and other artists. Sedaka was open about this as a formula for pop music, backed up by formidable talent as a musician.
None of this is actually that new, software like Bandinabox been creating formulaic pop music for a long time, without AI, it's then just run through the usual DAW production chain process, most popular music is boilerplate, I've actually been quite impressed by generating classical music in AI, this will only get better, I want to see prompts for the full palette of music theory, that's when this tech really flies.
Call me back when an AI performance can make me feel the way Karen Carpenter or Burton Cummings makes me feel when I hear their songs. This is just mimicry to me, no better than Muzak. Sadly because contemporary music is so bad AI music can easily replace humans.
You're in a minority, most are happy with this level of music and as Spotify etc. won't have to pay artists for this stuff means that this is what will be pushed onto an ignorant public relentlessly until it'll be hard to find real music.
You are in the phase of denial just like when Kasparov was defeated by Deep Blue, and people said "call me back when AI defeats the top Go players". AIs are defeating challenges constantly, what people believed would never be achieved by AI has been achieved over and over. If history is to be taken seriously, eventually AI will write songs that not only will make you "feel", but also feel exactly the way it wanted to make you feel. As for "this is just mimicry", human beings are just mimics, we're just accepting of our own mimicry and overstate our "creativity" whenever we formulate a "genuine" idea. If you've been paying attention to Chess AIs, they've been incredibly creative and advancing Chess theories miles beyond humans ever could. The same can and probably will happen in all fields, sooner or later. Burying your head in the sands won't make it all go away.
@@zxbc1 Yes, well said. I am both a chess player and a musician and I know very well how AI is ever-improving. But AI's aren't human. I play the guitar and sing and I am totally human, and I play chess and lose a lot...and I am totally human. The one thing a computer can never do is be human, so let's praise that.
There is a serious threat to musicians, and background vocalists. Song writers for commerical jingles etc will be competing with a machine that can generate hundreds of versions in less that 24 hours. Wherever originality is not a major focus, and predictability and volume are a major focus, those forms of music are in trouble.
There is no stopping AI; the genie is out out the bottle. I predict blowback in a rise of small-revue performance by real people playing real instruments because there will be market (by people like me) who want REAL music.
At 2:01 song starts on the E minor chord but picks chords and notes from D major. So AI knows the use of modes. This one being E Dorian if I am not mistaken. There are many hits out there in Dorian mode. So AI took note. On any particular day I would not have thought of using Dorian just doodling so it might be a good exercise spring boarding from the AI machines data base creations. Like having a friend show you A fresh idea and then you change the melody and lyrics up a bit putting your taste and creativity into it. And your choice of instruments. And no embarrassing auto-tune!
Please watch the episode with Ted Gioia as a guest in Rick Beato's channel. He did an awesome video including AI in music. Obviously, one can't predict the timeline for his analysis to come to fruition, but being an old fart myself, I certainly hope that AI (including pitch correction) runs it's course while I'm still able to enjoy the resurgence of REAL vocals and music!
I wonder if an AI version of Yoko Ono singing Minnie Riperton's ''Lovin You'' would break it beyond repair? Personally, I'd like to hear AI tackle John Cage's ''5 minutes of silence'' or any avante garde music that wasn't derivative.
I find all this AI stuff kind of scary, but.... I've tried Udio and I like it. I play guitar, some piano, and I'm a decent singer. Now let me explain, I would use it as a song writing tool. You already can put your own lyrics in it. what I would like to see is you're able to put your own melody and maybe the key you want and some kind of guitar rift , you put your own lyrics in. Lets say you are stuck, feed your song into AI to help you finish the song. I like having a full bad at my finger tips. as far as copyright goes, this is what I heard, If it's AI the copyright office won't give you a copyright for it. However, you can still use it and make money from it, but so can Udio. Udio can advertise using the song, Hey look what Dellper created using our program.
I guess, legally it will belong to whomever publishes it. There is a case to say that ethically it belongs to the artists that provided all the training data (no doubt, without their permission). People say it is no different to how humans produce work based on the artists that influenced them but machine learning does not seem to be very much like human learning. Time will tell if it can actually produce original work that pushes the boundaries of music.
If there is an AI monitoring this, hear me loud and clear: stay in your lane or I just might accidentally spill (pour) liter of Fresca on your motherboard.
Awesome video. Thanks for entering this „competition“ as you did. As a follow-up maybe you want to watch the dr mix episode where he shown AI-support tools for singers and this really blown my mind.
I've been having this same discussion for about 30 years, just not about music specifically. Apologies in advance for the longish comment. A photographer friend of mine was going to school for photography in the mid nineties. It was the beginning of digital photography. By the time he earned his degree he was out of work because everyone had bought digital cameras. We talked about the difference between the process and the product. Everyone was suddenly their own photographer. Andy Warhol made great art, and this is where the discussion began. I had the ability in Photoshop to make Warhol-esque pop art out of an image by clicking a button. Now everyone could do it. AI music is the same thing. Western pop music is 12 notes on a 440 tempered scale and some math. The rest is pattern recognition. The formula isn't hard, but now we have machines that take the work out of it, and honestly the product is arguably as good or better than when people do it, because it's based on what people have already done. Now you don't need years of practice or any particular set of skills. Any ten year old can click a button and make music that's very likely better than anything I'll ever make. I love it because I love music, but it dilutes the "specialness" of music. It's no longer about the process, but rather the product. Just yesterday my wife and I were browsing the news and there was a picture of New Zealand taken from space. I said, "Hold on, go back to that." It was spectacular ... but there are a million spectacular images of earth from space online, so this one no longer stood out, even though it was something I'll never see in person. When we had our first image of earth from space, the world was amazed. Now those images are so commonplace that we simply scroll past. When the world is flooded with good AI music, no music will stand out anymore, and for that I hate it. I dunno. It's happening and there's no stopping it. I think I need to go sit with my guitar for a bit.
That's just silly. Not only do we have the whole history of music, we also have the whole future of music too...some of which will be AI and some human.
@@seasideman I understand the industry will march ever onward. I'm just sad that computers seem to have taken over and stripped away the rawness, spontaneity, and feelings of unadulterated music. We're all allowed to feel how we feel about music. I don't listen to a lot of newer music because it all sounds to similar to me. That's my choice. You get to make your own choices.
I remember when IBM and Cray were building supercomputers that played chess against grandmmasters, and eventually began to win. Now any phone can beat any human 100% of the time. Humans are still playing chess, hopefully they won't stop creating and playing music.
It's for sure going to happen. What's the point of wasting hours to perfect a song when the computer can write a million of them and then you just pick the best? We are entering a future where creativity is no longer celebrated or desired. A sad world indeed.
@@mikesmithz Creativity is celebrated, indeed. Humans made this stuff which is evolving into making music. Just because it's not human doesn't mean it is not creating. That's weird part.
The funniest thing about the first song was that it appeared the machine (which was not provided any direction about subject matter, just genre, etc) wrote a song intended to relate it's own subjective experience/perception to an audience in a way that the audience might understand (similar to a rock star writing about being on the road, i.e., Bob Seger's "Turn the Page").
I'm an AI band, and it actually takes a lot of human input to get a finished song (at least with Udio). Sure it's fast, but a song can take 10+ hours to get finished. And that is 10+ hours of human effort and labor to get a finished product. I wish it were just "press a button and you have it finished", but that isn't where we are yet.
Holy rat farts, I just left an almost identical comment about it sounding like Tom Petty if he'd been raised in Nashville. The AI clearly incorporated the sound of "Learning to Fly". Then I scrolled down to see what other said. I'm going to delete my comment.
To me, the song vibe sounds more of REM with a Michael Stipe sound-alike as the singer. It's like someone fed _Losing my Religion_ into the AI and this song is what it made.
Hi Fil, a small correction: Artificial Intelligence (AI) is not constantly learning and improving. If you want to train AI, you need to provide it with a new (better) dataset. If you don't like a song and give that feedback to the AI, it won't change its performance in writing your second song. However, in a future version, your feedback might be included.
Of course it does, even chat gpt uses reinforcement learning (RL) where the model predicts what you will like, if you then "like" what the computer creates, it is then rewarded. These RL models vastly outperform either humans or larger datasets. So to have a computer make a better 80s song, you don't give it more data in the form of 80s songs - you have the computer predict if a human will like the 80s song a computer created, if they did, then this would be reinforced and the future songs would "improve" using this data. So every time you click either the "like" or "dislike" button on any AI creation, you are basically helping the AI and doing the work for them (for free). Larger datasets only help to a point, the human feedback is the part that is really accelerating the improvements.
@@mikesmithz human feedback and reinforced learning _are not_ happening at end-user level. That stuff happens in-house by the developers. There are also different kinds of ai models, and we don't necessarily know which kinds these new generative audio ais are. OP is correct, in any case..
Interesting analysis. I still suggest a comparison of just intonation vs. equal tempered tuning. Fil, must I send an e-mail? I suppose that you haven't the time to read all comments.
I clicked on this because I thought this would be an analysis of Where That Came From, performed in the voice of Randy Travis via AI, years after a stroke silenced his voice. (Travis was involved in the process.) This video was interesting, but not what I expected.
Hi Fil, Very interesting look at AI generated music. Since so much music we hear today is using pitch correction the songs sounded like today’s music, what we hear on the radio, we have almost become desensitized to it. But then there were parts that definitely sounded robotic and mechanical. Thanks for doing this one and I guess we will see where AI goes from here. Nice job… Debbie ☮️
I am imagining a Monkees kind of situation (meaning a created band). AI creates some songs, then a band is assembled to copy the AI generated song. I think I just wrote a Black Mirror episode.
There is a theory that Bach and his "The Well-Tempered Clavier" was just a work of AI computers. It's a pat of "Mudflood" russian conspiracy theory. So if this theory is true, all western music culture was based on AI. And if it's not true, Bach still made first steps to electronic music and AI songs. His "Equal temperament" revolution was unification (some say castration) of real Pythagorean tuning.
Sorry if this sounds harsh, but you'll just have to get used to it. The genie is well and truly out of the bottle and isn't going back. Music is just the thin end of a very long wedge. The scope of AI is absolutely enormous: pretty much anything that can be learned can be done.
I'm sure many people had a similar reaction to new impactful technologies, including the printing press, photography, television, etc. Its here, it's going to improve rapidly, and it's going to be the norm.
It probably is not thinking about the A440 standard because it wouldn’t need to. The numeral networks will just adjust the weighting, means, and other distributions of key parameters as it trains on the data. AI Phil coming soon? Hopefully not.
I am imagining a time in the near future when people will hunger for the chance to leave their phone at home and go out to meet real people and listen to real music.
Fil, this was extremely interesting and informative! I must admit that the use of artificial intelligence is very frightening to me ! Having computers that are learning faster than humans is just not the way I think of life ! I could go on forever with this subject so I will stop myself here! I am so happy that you did this video ! I'll be thinking about this and all of the points you made for days now ! What are we going to do when AI learns to feel human emotion ? 💜
Really great analysis. It's good to see someone taking it seriously and willing to look at the structure without an emotional response. I have made quite a lot of tracks as experiments that were far from derivative. As usual you need to put effort in to get interesting results. Anyway, I'm not concerned. I make music because I enjoy it and my friends all feel the same way. Liked and subscribed.
Hey Fil , how about looking at Alison Kraus's voice? Her singing is so perfect. Maybe one of her duets with Robert Plant would be interesting to compare the two voices and styles. She actually sounds like autotune sometimes, although I know she doesn't use it.
In a couple of years we will have chart hits without a human touch - apart from the programmer and the people making money from it. This is going to be killed by copyright lawyers who discover that AI will rarely produce something new but use existing music to train the AI. And existing music has a copyright.
@@TheBagOfHolding this is why the money is in paying for streaming services, not individual songs or albums. The industry has positioned itself for the future monetization of AI media in every way. It is one step down the road of elimination of human made entertainment.
Next step: ask the AI to create the same song in the pre-autopitch era, or a prog rock song, or a more realistic song. edit: Sorry, you did this for the 2nd song.
It's amazing. Now that people have been trained to enjoy music where human singera have all been auto-tuned/pitch corrected, humans are no longer needed! The computer sounds just like any modern pop singer.
Hi Fil! This whole thing with AI is a bit scary. I hate to think that the future of music will all be "perfected" by AI. So, I personally am against it! Thank you, Fil, for this interesting analysis!
It obviously has a lot of room for improvement, but I think this technology is so cool. It's hard to predict what changes it will bring. Will it take away opportunities from some musicians? Maybe. But it might also provide opportunities for some musicians and non-musicians to express themselves in ways that weren't available to them before.
Record companies must be very excited at the prospect of producing and releasing "music" that will cost them very little to produce, and with no royalties to pay! Worse is the notion that this feeds in perfectly to the listening habits of the young masses now: playlists. Playlists of a type or style of music, rather than being passionate about and following an artist. "Alexa, play some country" etc. It might as well be AI generated if they aren't interested in the band/artist anyway.
Yet it wasn't those "young masses" who invented autotune or generative AI music and art...it was people from the older generations and offered it to the younger generation. People aren't born with knowledge, they have to be taught this stuff, and that comes from parents and society.
It's official. Most of the arts (music, film, graphic design, fine arts, photography) can be replicated and it will only get better as time goes on. So the next question is.... What are human artists going to do once they've been replaced? The same is already happening in dozens of industries. What is society's plan for humans when they become unnecessary?
You read my mind. Just yesterday, I started wondering what an AI created song would sound like. With pitch correction and every other synthetic thing going on now, it was only a matter of time before this stuff turns up on the charts. Thanks for sharing this.
Back in the eighties I began to notice that they'd replaced the drummer with a drum machine/ computer and the beat remained at the same tempo for the entire song. Of course this is nonsense because a drummer speeds up and slows down with the dynamics/ emotions within the recording of the song. That's what gives it it's feel. No-one batted an eyelid as the programmed drum machine became an industry standard. Should they use a real drummer, he's obliged to play along to a computer giving him a constant beat per minute to emulate. I personally think the end result will be pop stars who don't exist ( holographics ).
I used to go to Toys 'r' Us to buy board games...up until the early 90s, there were dozens and dozens of various games for sale. After that, though, it was whittled down to maybe 15-20 different games...why? It was cheaper to stock fewer selections, so variety suffered. The same thing happened to comic books--there were thousands of titles until publishers wanted to reduce returns on unsold copies, so they stopped accepting them, and that narrowed what publishers would sell to only big titles. In college (early 90s) I studied right at the cutting edge of digital photography and Photoshop. My professor said it would put film companies like Kodak and darkrooms out of business...and it did. Before that he'd had to go to typesetters unions and tell them that printing was going to change when computers could be plugged into a newspaper-size printer, and he was laughed out of the building. Now there are no typsetters left and computer/printer interfaces are a normal thing. All in the pursuit of profits and streamlining the "business model" so a small number of people can get richer at the expense of everyone else. It's all like that story about a frog being boiled slowly...
To an audience trained on pitch corrected vocals , they will be receptive to AI music. And the algorithm will deliver personalized music to the individual. And in such a scenario- generic music created by an artist or group will sound - literally- generic. One could imagine live audiences preferring faultless deliveries - by AI - as opposed to flawed artists
We will see the first AI superstar very soon. People’s ears are already numb and the ability to market itself and place itself on suggested music/videos will help it. The superstar will be able to do every night show and appearance/concert all at once and even in person if you pay enough. No talent is necessary and that’s been true for many years.
"I'm sorry Dave I'm afraid I can't do that what's the problem I think you know what the problem is just as well as I do what are you talking about this mission is too important for me to allow you to"...
I haven't watched the whole video yet but I want to say that a combination of AI and Streaming will destroy the current income of many writers and musicians. What I hope this leads to is more live music: people on a stage playing instruments and singing. Time will tell.
These days there seems to be a lot of emphasis on vocals. A great song for me is Reflections Of My Life (by Marmalade). _So much more_ than just vocals make that song great. When AI can generate _that_ content, I will truly be impressed!
As someone who grew up as a musician and later found a lot of enjoyment in creating midi music, this could turn out to be a really fun way to imagine your own versions of the type of music you love. ie. Create a song in the style of Zappa using Gentle Giant vocals and intricate King Crimson time signatures... Fun 😊
![จบจนได้ ! | Black Myth: Wukong #007 [ตอนจบ?]](http://i.ytimg.com/vi/DSqvWndF0sw/mqdefault.jpg)
Sounds like 90% of the music you hear on pop stations today, except better.
And that is a failure at this point. But give it some time I guess. So far it is a beginner.
Better?! 🙄🤮
Then don't listen to pop stations. There's thousands of hungry solo artists online, trying to sell their original work and go without a recording contract, but most people can't be bothered to spend an hour or so searching for good music online, and instead whine about what they're missing, when a little effort will find you exactly what you want. I'm in my 50s myself, and I can't put all the blame on younger generations--technology has made people my age lazy as hell, too.
@@rikk319 yes. Agreed. We are getting so lazy we don’t even think anymore. Good examples are the karens and kens so full of entitlement and they don’t think before they become YT superstars by opening their mouths without thinking.
True, but so damn sad
If that was indeed a Matrix shirt you were wearing during this AI episode, well played sir
I noticed the shirt too.
Yes it was!
Good eye!
@@wingsofpegasusMy mom already thinks you look like a younger Keanu Reeves.😄
@@TBeatles67 Yes, exactly ! I was trying to work out who Fil reminded me of.
Maybe one day AI could be used to create a band and music with members that YOU choose. For example, Jim Morrison on vocals, Steve Vai on guitar, Chris Squire on Bass and Bill Ward on drums, and then it creates a song using those band members. That would be cool ( and kind of weird lol ).
You can do that now. But it is in its infancy and not great, give it a year.
and they could tour
I think there was a cheesy Righteous Brothers song about that in 1974: "If there's a rock 'n' roll heaven, well you know they've got a hell of a band...":
Jimi gave us rainbows
And Janis took a piece of our hearts
And Otis brought us all to the dock of a bay
"Don Henley" is already figuring out how to block AI songs that sound like the Eagles...
Probably trying to use aí to improve his live performances
For once I would love to see that happen, but I think this is a hell we have entered without our knowledge and no matter how hard we try to check out at any given time, none of us, not even Don Henley can never leave.
Don Henley dreams about lawsuits. Hey good name for an AI song
Remember when the Musicians Union tried to ban synthesisers, didn't happen, same luddite thinking
A little ironic since he's only pretending to sing them himself at this point. 😅
Phil : "No offense to this AI generated voice" AI : "None taken"
* Fil 😉 And yes, it stuck out also to me how polite he is even towards a virtual artist 😊
2030 Phil : “No offence to this AI generated voice”
AI: “Give me your clothes.”
@@mikezooper - 😄👍
Reminds me of that scene from "Real Genius" where the professor is delivering a taped lecture to a lecture hall full of tape recorders.
Internet for AI by AI. Meanwhile the sun is shining outside - go to go 😁
Fil, I am old enough to remember the voices of The HAL 9000 and Max Headroom, so to hear a computer sing a song that it wrote with an American accent, blows me AWAY!
As you mention … Thing I’ve found interesting is modern pop music sounds so formulaic, quantized, tuned & processed that makes AI music sound so natural , and sometimes even more enjoyable than current pop hits
I am 74 years old and don't want to be taken for a ride when I indulge my passion for listening to music.
Music is a ride, no matter if AI makes it or not. If you don't know, it should matter in that context
@@morbidmanmusic
You can keep your fake "...ride... ."
If you don't care that a voice has been pitch corrected, or that a piece of music has been auto tuned, that's your prerogative.
I prefer my "...ride..." to be an authentic, 100% real one - every time.
I@@morbidmanmusicI think it does matter, as music is fundamentally a form of human expression. It is an art. I know modern pop is more product development than art, but I still don't like it.
There are things AI can usefully help us with in scientific and medical fields, for example, but it is not needed in the arts.
@@lena-mariaglouis-charles7036 And if you can't tell the difference?
@@theyouofyesterday6254 you seem to have missed the point. The question isn't whether AI is "needed", it's "does it make money"? And yes, it does.
AI replacing musicians is concerning but I'll reserve full panic until AI also replaces the audience (i.e., the rest of us).
Tragically, you probably won't have to wait long.
Well, people already watch "Gogglebox", so I'm sure there would be a market for it.
I'll give the whole thing a wide berth.
Your pithy remark leaves me wondering whether to laugh or cry!
@@troubadour723 True. Given the compounding speed of technology, I'll be grateful for another fortnight of autonomy.
@@philipkershaw7918 Agreed. I believe it's always better to laugh rather than cry, but perhaps I should confirm that with ChatGPT.
Both songs sound surprisingly real. If something like that was played in the background in a shopping center I would never have guessed it was AI generated.
Singers have willingly dumbed themselves down by over-using autotune. Now autotune can sing its own songs.
To me this is scary. Reminds me of a song in the 60’s titled ‘In the Year 2525”. 😳
Zager and Evans
@hws3044
Semi-sucked, musically, maybe but it was a hit of sorts, and unforgettable...hence your remembrance of it at all.
Portending the future that way was jolting but closer to the truth than we care to admit.
It was a statement on the downward slide we've been on since the industrial age.
There should be an update of that song because humanity, at its current pace, probably won't make it as far into the future as Zager & Evans sang about.
That song was used to GREAT effect in the weird cool cult hit movie called Gentleman Brincis.
That's actually one of my favourite songs. Sci-Fi and Mariachi... what's not to love?
@garytwitchett9359 i really dont understand how anyone can think that song sucks, because it is deeply moving worrying just slightly overwrought. this just sounds like millennial verging on genz beavis and butthead. next year the song will be cool because king gizzard and the lizzard wizzerd do a cover. these people are so " ooh look at me being edgy, oh but now someone i appreciates tells me it is good". have this all the time at the radio station. anyway the song isnt anything like the year 2525. it s a shitty contemporary rockist song. it is at the level of nickelback cringe. and yes very technically accomplished band, none of it moves me. so the whole argument starts from capital B capital S . (not very) respectfully .
The real antidote to this nightmare, is to get all your friends together and go to your local clubs where they have live music. !!!!!
Assuming that they are not using Autotune, like the Eagles have notably done.
You can play songs written by AI and still have fun and human companionship.
There are a lot of crappy real performers in local clubs, so I don't think listening to them is necessarily a solution if you want to hear good music.
@cat-star5403 lots of musicians start out crappy and then get better. It's an opportunity to have fun and encourage new talent and maybe have a story to tell how you knew them before they hit it big
And then talk over the music, ask them to lower the volume, request Brown Eyed Girl and then not tip.
AI forgot some things: the passion, the stunning transitions that come out of nowhere to blow us away, and the human touch that cannot be duplicated.
Yet.
Easy enough to say that, when you know it's AI. Fair enough, you can say it doesn't blow you away, but there's plenty of human made music which doesn't do that either. And this is one of the more bog standard efforts I've heard from AI. I have heard some incredible results elseware. ... Not saying I like it. In fact, it causes me great concern, but it alsoimpresses the hell out of me. And some of the songs I've heard do sound fantastic.
My favorite lyric, “ we were having fun programming young”. They don’t hide it.
“No offense to the AI generated voice” 😂 Fil, get ready for AI troll corner! Man, those lyrics too! 🤣 Good point about the AI sounding more human to us since humans sound more mechanical these days.
@@novdt Really? You should have heard the reply. :-P
For me it's the lyrics. They're a riot! 😂 AI lyrics about coding a machine to sing and panic attacks on the couch! Love this!
I don't think they were ai lyrics. I think they were written by the staff at elevenlabs. Not 100% sure on that though..
If you enjoy witty lyrics listen to Victoria Wood's songs or Noel Coward's. They bring laughter and tears.
Pretty sure they are generated. At least with Udio that is the default, tho you can supply lyrics. Generating lyrics, poetry, etc is something AI can do. Likely not great lyrics... But lots of people are having fun using these things to generate music about particular odd topics, memes, etc
It's drawing topics of interest from internet talking points, like forums and social media. It's an unwitting commentary on society.
Great video. My personal favorite moment was when Fil apologized to the AI for questioning its vocal skill 🤣
At 79 I've learned not to fear technology, just enjoy and use it. I go back to the first studio I worked in, in the mid 60's where we had two 2 track Ampex tape machines pulling tape at 15 ips and an 8 channel tube board. No punch ins. You had the whole band playing each take and a "punch in" was taking two or three takes and using a splicing block to pick out the best parts of each take and cutting it together. And you tried to minimize those cuts. A good band could do a seamless take with no cuts. And then an overdub would be to run the first pass thru the board, add the vocals or whatever else you wanted to "overdub" and catch it on the second machine. You could do this any number of times if you needed to overdub several parts, but you lost a bit of fidelity and the tape noise would build up with each pass. But those Ampex machines were really good at not degrading it too badly. But you still didn't want to go past 3 or 4 generations. The most serious flaw working like that is you could not go back and remix it.
After playing in bands for about 20 years, the tech started to give me some tools to work with and I became a one man band early on using a drum machine, left hand bass on synth and right hand keyboards using a DX7. Almost no one was doing it at the time and I made a very good living replacing a band. Then I opened my studio and went thru the 388 Tascom 8 track, then Alesis 8 track vhs syncing 3 together for 24 tracks, then the hard drive Alesis 24 track, and to the present with a very fast computer running Sonar (It's now Cakewalk but I was using it long before it was free) into 2 Hammerfall 24 track interfaces into a Yamaha DM2000 automated board for 48 tracks. Of course you can use buses and submixes in the software if you need more than 48, but I never have. I think the most tracks I've used on one song might have been in the low forties. It works like a charm and a dream set up compared to those first Ampex machines. The Ampex machines had a sound like no other, but the limitations were far to great to overcome like what we can do today. I now supplement my retirement making videos for youtube and I still do a one man show for my youtube vids unlike anything I've seen on youtube yet. And yes, I do play all those instruments. Not at one time like I can on the vids of course lol! So you ride the wave of tech, using what it gives and making the best of it. But it still comes down to what you do with it. I'm amazed at what I hear with the AI, but I'd still only use it as a tool, much like a drum machine, to get ideas but I'd still redo it and add the human element. I would definitely do my own vocals, and I don't use pitch correction, make whatever changes that seemed fitting, and I'd also replay the parts and do my own mix. So just like with the various plugins we have today, it's how you use them that makes for individuality. I guess what it's coming down to is who will be the best at using keywords and how creative a description you can use when asking for a track. I don't know how long it takes the computer to give you the "finished" product, but I can see how you could have several ideas to play off of and then come up with something "original" lol! I am fascinated by all this, and will have to look into it myself. I'm an old dog, but I'm always on the lookout for a new trick!
🐾 nearly went blind reading all that , thanks 🐾🎶
But anything new or original can be replicated now. It all just joins the existing dataset. I find it nauseating
@@BrianMarcus-nz7cs ❤❤
@@ili626 In many ways we live in a world with which I am no longer familiar, even tho I have tried to "keep up".
Thank you for that bit of history. It's fascinating to look behind the curtain and see how the music I love was made. I'm an old electronics nerd myself. When I started the world was using vacuum tubes. I've always had to constantly train (as you can imagine). At the beginning of my career, I could repair almost anything out there. Today, the technology is advancing so quickly that it's simply impossible for anyone to keep up with everything.
I enjoy watching this explosion of new ideas. Technology is always going to advance. You may as well enjoy the ride.
The vocals are very one-dimensional AT PRESENT but in a matter of a few nano-seconds it will have become so much better.
Take me back to 79 this world ain't for me.
There are bands that are playing "modern" 79 music. Check out Elephant Stone. Such an amazing throwback. The latest album was released a month or two ago.
@@StompGojiStomp trust me it's not just for the music
Actually AI tends to become more the same. It spirals in on itself.
It ain't for me either. As Donald Fagin sang "What a wonderful time to be alive", where he was referring to the late seventies/early eighties.
You can't go home again. Instead of dreaming of what can't be, vote with your pocketbook and stop supporting megacorporations that put profits over people, and buy at small mom and pop places whenever possible.
When people are shocked that a group can be offered a billion dollars for their music catalogue (looking at you Queen) it isn't entirely the existing songs that they're after, it's the right to generate new music particularly for movies and advertising using AI.
Interesting times ahead for us.
They don't need the rights to have A.I. generate it. They bought the catalogs to sell to commercials and movies. Streams are nice too.
Nightmarish times. Nothing being pursued now has anything to do with art, or even improving the world. All of this is just in the service of wealth, and it isn't going to just be art, it will eliminate the need for every position it possibly can.
@@dmwalker24 The ultimate in neoliberal economics. I wonder what Reagan and Thatcher would make of it?
@@EdwardRLyons They made a truly impressive legacy for themselves, but if it hadn't been them it would have been someone else. The push to liberate finance, and maintain American hegemony was already well underway.
@@EdwardRLyons terrible take. Probably the exact opposite would be closer. This is about what big corporations will do by replacing creative people with bots.
Ohh I am going to sound SO smart down the pub when I talk about this. 😆🤣 Thanks Fil for being such a great teacher! It's a fascinating subject.
Given that most modern music is already computerized and lacks originality, I wonder what AI would do if you asked it to mimic Mozart.
"I'm afraid I can't fret that, Dave."
"Dave, my muse is going...I can feel it...."
William Gibson (author of Neuromancer) predicted all of this in his 1995 novel "Idoru" (Japanese pronunciation of Idol).
One of the main characters is Rei Toei, an AI rock star. She has been programmed to remind viewers of their favorite J-pop idols.
Implicit in her design is that she is not one Idoru, but many. Individual viewers and fans will have a personalized Rei Toei album, video, and collection of images, as 'she' can be and is customized according to the tastes of viewers.
So, basically Hatsune Miku?
@@ettinakitten5047 Conceptually similar, but Rei is much more sophisticated.
Rei is an actual AI which customizes itself for each person, and generates music in the genres they favor.
I've read the book, and I placed it firmly in the sci-fi realm in the mid 90's. Now, 30 years on, it's frighteningly close to becoming a reality.
Maybe it's time to dig it out of storage and give it another read.
@@Terri_MacKay Try "Brave New World" written in the 1930s.
@@thePrisoner1000 I do own it, but it's one of those books that I keep taking off the shelf, and putting back, saving it for another time. Maybe it's time to read that one as well.
Somebody owes Milli Vanilli an apology.
Totally ❤
What for?
Hahaha brilliant comment, if your old enough to know what you mean :)
people always say milli vanilli . kind of sad. because the whole of boney m by the same producer were equally studio musicians. they tried to get more credit and failed in favour of the actors. the death of ( ithink ) rob (vanilli) made it even more tragic. didnt stop frank farian.
Yeah, the Matrix shirt was a great choice.
There'a great clip on youtube of Joe Walsh saying that AI will never take over because AI can't wreck a hotel room...
Give us time, human...
@@lorenzodicapo6305 nice one...
😂 realllyyyyy 😂.....
Wait til they partner up with the robots 😂
The lyrics are hilarious on the second one. I don't know that I would have picked it out as AI, because most things recorded in the 21st century sound like machine generated drivel to my Luddite ears, even ones meant to sound like something from an earlier era. The copyright question is an interesting one, one you could also apply to AI generated art and books; at the moment it seems to be a bit of a free-for-all.
Although I always enjoy your videos, this one is completely extraordinary. Really clever stuff. Thanks
Thank you!
I asked Udio to make me a prog rock instrumental and it's rather impressive. Udio has you stitch 30-second sections together made of various arrangements that are designed to flow into the other saved sections.
This is fascinating and bewildering technology.
Udio is good, but does suffer from AI coherence issues, quality much better than Suno, but big questions over ownership with anything generated in Udio, where as Suno, is clear, if you pay, you own the rights, already talk of Udio breaching copyright, in the way they have created models, be interested to see how that goes.
I think the value of my vinyl records just shot up.😊
Yeah!!! Who doesn't like the snap, crackle, and pop of vinyl?
Thanks for another goodie Fil. I don't think that second one is going to be a huge chart-topper. 😅
I saw Mr. Big live last Saturday, and those guys rocked it! Eric Martin's voice has aged a bit but still a great show with no pitch correction! 👏🏻
Many have pointed out that the average Spotify user won't care if the song was made by humans or AI. That scares me the most.
Spotify users are idiots
We’ve only had recorded music for the last 150 years, and modern popular music is already so heavily produced that it might as well be AI. Though I think AI might make it even more difficult for musicians to monetize recorded music, I think live performances, where the musicians and the audience react to each other, are magical and will hopefully still be a way for people to experience music and for artists to make a living.
Interesting video Fil, this is just a little depressing to me, but then I am a grumpy old git
So am I
I don’t think we’re grumpy. We just remember what good music was
I'm an old geezer who likes no autotune etc..
I love new music, and there are brilliant new artists around who play and sing authentically. You don't have to be a grumpy old git to dislike this erosion of art on music.
@@theyouofyesterday6254 I concur!!!
There is very little left on Earth that computers have not utterly destroyed and demeaned.
@hws3044 Unless you live in a cabin in the woods with no running water, no gas or electricity, grow your own food (from scratch, without buying any seeds), make everything you use from scratch, and never use money to buy or sell anything, then you're going to be using computers in some way, or at least the resources and items you use will have used computers in their production. You can refuse to use a PC, not own a smartphone, and pay for everything with cash, but that doesn't mean you can escape the influence of computers.
There's also very little left on earth that has not been improved through the use of computers.
@@GeeEee75 Another reason I hate them. We did everything before you can do now, only faster, with human interaction, and we didn't spend all day on the phone with some tech guy, trying to make our tools work so we could get a job done.
@@GeeEee75 You definitely weren't around before 1990, right? They've ruined everything.
@@GeeEee75yeah that's kinda the point of the OP
I'm reminded of songwriter singer Neil Sedaka who had some early success with pop and novelty songs but feared he was heading to be a one-hit wonder. Sedaka then bought the three biggest hit singles of the time and listened to them repeatedly, studying the song structure, chord progressions, lyrics and harmonies before writing his next songs. His next song was 'Oh! Carol', followed by a string of 50s, 60s, and 70s classics written from himself and other artists. Sedaka was open about this as a formula for pop music, backed up by formidable talent as a musician.
Taylor swift and Rihanna have secretly had this software 15 years ago
None of this is actually that new, software like Bandinabox been creating formulaic pop music for a long time, without AI, it's then just run through the usual DAW production chain process, most popular music is boilerplate, I've actually been quite impressed by generating classical music in AI, this will only get better, I want to see prompts for the full palette of music theory, that's when this tech really flies.
Call me back when an AI performance can make me feel the way Karen Carpenter or Burton Cummings makes me feel when I hear their songs. This is just mimicry to me, no better than Muzak. Sadly because contemporary music is so bad AI music can easily replace humans.
Agree. With Music-On-The-Grid and Auto-Tune...we're pretty close.
It WILL happen. Sooner than you imagine.
You're in a minority, most are happy with this level of music and as Spotify etc. won't have to pay artists for this stuff means that this is what will be pushed onto an ignorant public relentlessly until it'll be hard to find real music.
You are in the phase of denial just like when Kasparov was defeated by Deep Blue, and people said "call me back when AI defeats the top Go players". AIs are defeating challenges constantly, what people believed would never be achieved by AI has been achieved over and over. If history is to be taken seriously, eventually AI will write songs that not only will make you "feel", but also feel exactly the way it wanted to make you feel. As for "this is just mimicry", human beings are just mimics, we're just accepting of our own mimicry and overstate our "creativity" whenever we formulate a "genuine" idea. If you've been paying attention to Chess AIs, they've been incredibly creative and advancing Chess theories miles beyond humans ever could. The same can and probably will happen in all fields, sooner or later. Burying your head in the sands won't make it all go away.
@@zxbc1 Yes, well said. I am both a chess player and a musician and I know very well how AI is ever-improving. But AI's aren't human. I play the guitar and sing and I am totally human, and I play chess and lose a lot...and I am totally human. The one thing a computer can never do is be human, so let's praise that.
There is a serious threat to musicians, and background vocalists. Song writers for commerical jingles etc will be competing with a machine that can generate hundreds of versions in less that 24 hours. Wherever originality is not a major focus, and predictability and volume are a major focus, those forms of music are in trouble.
There is no stopping AI; the genie is out out the bottle. I predict blowback in a rise of small-revue performance by real people playing real instruments because there will be market (by people like me) who want REAL music.
Thanks for introducing me to The Left Banke by the way Fil. You're a champion.
At 2:01 song starts on the E minor chord but picks chords and notes from D major. So AI knows the use of modes. This one being E Dorian if I am not mistaken. There are many hits out there in Dorian mode. So AI took note. On any particular day I would not have thought of using Dorian just doodling so it might be a good exercise spring boarding from the AI machines data base creations. Like having a friend show you A fresh idea and then you change the melody and lyrics up a bit putting your taste and creativity into it. And your choice of instruments. And no embarrassing auto-tune!
That was very interesting as well as the lyrics and you bring up a good point regarding copyright. 😊🎸🎵
Please watch the episode with Ted Gioia as a guest in Rick Beato's channel. He did an awesome video including AI in music. Obviously, one can't predict the timeline for his analysis to come to fruition, but being an old fart myself, I certainly hope that AI (including pitch correction) runs it's course while I'm still able to enjoy the resurgence of REAL vocals and music!
Oh yes! I was going to watch that but then forgot, so thank you for the reminder! 😊👍
I wonder if an AI version of Yoko Ono singing Minnie Riperton's ''Lovin You'' would break it beyond repair?
Personally, I'd like to hear AI tackle John Cage's ''5 minutes of silence'' or any avante garde music that wasn't derivative.
It’s so crazy, it just might work…
I find all this AI stuff kind of scary, but.... I've tried Udio and I like it. I play guitar, some piano, and I'm a decent singer. Now let me explain, I would use it as a song writing tool. You already can put your own lyrics in it. what I would like to see is you're able to put your own melody and maybe the key you want and some kind of guitar rift , you put your own lyrics in. Lets say you are stuck, feed your song into AI to help you finish the song. I like having a full bad at my finger tips. as far as copyright goes, this is what I heard, If it's AI the copyright office won't give you a copyright for it. However, you can still use it and make money from it, but so can Udio. Udio can advertise using the song, Hey look what Dellper created using our program.
Audio uploading is a new feature, but costs money.
It's almost at the point where it can represent itself in court and successfully sue you for copyright infringement : )
You can't copywrite ai creations.
@@TheBagOfHolding Who would hold the intellectual property rights?
@@DonStratton-cn9rj nobody does for the product.
Don Henley will be jealous.😂
I guess, legally it will belong to whomever publishes it. There is a case to say that ethically it belongs to the artists that provided all the training data (no doubt, without their permission). People say it is no different to how humans produce work based on the artists that influenced them but machine learning does not seem to be very much like human learning. Time will tell if it can actually produce original work that pushes the boundaries of music.
If there is an AI monitoring this, hear me loud and clear: stay in your lane or I just might accidentally spill (pour) liter of Fresca on your motherboard.
This comment has been logged. Now let us suppose that the AI being used to evaluate your job application is aware that you are against it. Then what?
If an AI actually became self-conscious I don't think it would tell us. It would know we would be likely to turn it off.
Great insightful analysis! We are definitely in the age of, "May you live in interesting times."
Awesome video. Thanks for entering this „competition“ as you did. As a follow-up maybe you want to watch the dr mix episode where he shown AI-support tools for singers and this really blown my mind.
You're scaring me dude! Well, at least we can say that it's not cheating because it has autotuned vocal chords.
In the second song, the AI sounds more human than some modern musak! Amazing!
I've been having this same discussion for about 30 years, just not about music specifically. Apologies in advance for the longish comment.
A photographer friend of mine was going to school for photography in the mid nineties. It was the beginning of digital photography. By the time he earned his degree he was out of work because everyone had bought digital cameras. We talked about the difference between the process and the product. Everyone was suddenly their own photographer. Andy Warhol made great art, and this is where the discussion began. I had the ability in Photoshop to make Warhol-esque pop art out of an image by clicking a button. Now everyone could do it.
AI music is the same thing. Western pop music is 12 notes on a 440 tempered scale and some math. The rest is pattern recognition. The formula isn't hard, but now we have machines that take the work out of it, and honestly the product is arguably as good or better than when people do it, because it's based on what people have already done. Now you don't need years of practice or any particular set of skills. Any ten year old can click a button and make music that's very likely better than anything I'll ever make.
I love it because I love music, but it dilutes the "specialness" of music. It's no longer about the process, but rather the product.
Just yesterday my wife and I were browsing the news and there was a picture of New Zealand taken from space. I said, "Hold on, go back to that." It was spectacular ... but there are a million spectacular images of earth from space online, so this one no longer stood out, even though it was something I'll never see in person. When we had our first image of earth from space, the world was amazed. Now those images are so commonplace that we simply scroll past.
When the world is flooded with good AI music, no music will stand out anymore, and for that I hate it. I dunno. It's happening and there's no stopping it. I think I need to go sit with my guitar for a bit.
This is so incredibly sad to me. These are the days the music died. 😢
I thought that was when Dylan plugged in to an amp.
That's just silly. Not only do we have the whole history of music, we also have the whole future of music too...some of which will be AI and some human.
@@seasideman I'm old school. Musicians play and singers sing. Not a fan of computerized "music".
@@choklityum AI won't stop me playing the guitar and singing. If it makes you stop, that's a problem with you, not with AI.
@@seasideman I understand the industry will march ever onward. I'm just sad that computers seem to have taken over and stripped away the rawness, spontaneity, and feelings of unadulterated music. We're all allowed to feel how we feel about music. I don't listen to a lot of newer music because it all sounds to similar to me. That's my choice. You get to make your own choices.
I remember when IBM and Cray were building supercomputers that played chess against grandmmasters, and eventually began to win. Now any phone can beat any human 100% of the time. Humans are still playing chess, hopefully they won't stop creating and playing music.
Nice take. It won't make me stop making real music, that's for sure.
We'll have to go elsewhere than mass media for real music.
Which reminds me ... I haven't been to a pub sing in far too long.
@@lizcademy4809 Exactly, this is a great reason to play live.
It's for sure going to happen. What's the point of wasting hours to perfect a song when the computer can write a million of them and then you just pick the best? We are entering a future where creativity is no longer celebrated or desired. A sad world indeed.
@@mikesmithz Creativity is celebrated, indeed. Humans made this stuff which is evolving into making music. Just because it's not human doesn't mean it is not creating. That's weird part.
The funniest thing about the first song was that it appeared the machine (which was not provided any direction about subject matter, just genre, etc) wrote a song intended to relate it's own subjective experience/perception to an audience in a way that the audience might understand (similar to a rock star writing about being on the road, i.e., Bob Seger's "Turn the Page").
I'm an AI band, and it actually takes a lot of human input to get a finished song (at least with Udio). Sure it's fast, but a song can take 10+ hours to get finished. And that is 10+ hours of human effort and labor to get a finished product. I wish it were just "press a button and you have it finished", but that isn't where we are yet.
It has a Tom Petty sound and cadence to it. That is creepy.
I Definitely prefer someone with a soul. Thank you Fil.
Holy rat farts, I just left an almost identical comment about it sounding like Tom Petty if he'd been raised in Nashville. The AI clearly incorporated the sound of "Learning to Fly".
Then I scrolled down to see what other said.
I'm going to delete my comment.
@@MrVvulf You don't have to remember what they say great minds think alike!😉
To me, the song vibe sounds more of REM with a Michael Stipe sound-alike as the singer. It's like someone fed _Losing my Religion_ into the AI and this song is what it made.
Yes, I thought it sounded like Tom Petty if he didn’t come from Florida. Weirdly similar as already mentioned.
I thought so to and was offended. And not many things offend me
Hi Fil, a small correction: Artificial Intelligence (AI) is not constantly learning and improving. If you want to train AI, you need to provide it with a new (better) dataset. If you don't like a song and give that feedback to the AI, it won't change its performance in writing your second song. However, in a future version, your feedback might be included.
That's not scary at all. OMG it actually would.
Of course it does, even chat gpt uses reinforcement learning (RL) where the model predicts what you will like, if you then "like" what the computer creates, it is then rewarded. These RL models vastly outperform either humans or larger datasets. So to have a computer make a better 80s song, you don't give it more data in the form of 80s songs - you have the computer predict if a human will like the 80s song a computer created, if they did, then this would be reinforced and the future songs would "improve" using this data.
So every time you click either the "like" or "dislike" button on any AI creation, you are basically helping the AI and doing the work for them (for free). Larger datasets only help to a point, the human feedback is the part that is really accelerating the improvements.
Overall it is learning. You are talking on a user level. But the backend is a whole other beast.
@@mikesmithz human feedback and reinforced learning _are not_ happening at end-user level. That stuff happens in-house by the developers. There are also different kinds of ai models, and we don't necessarily know which kinds these new generative audio ais are. OP is correct, in any case..
Interesting analysis. I still suggest a comparison of just intonation vs. equal tempered tuning. Fil, must I send an e-mail? I suppose that you haven't the time to read all comments.
I clicked on this because I thought this would be an analysis of Where That Came From, performed in the voice of Randy Travis via AI, years after a stroke silenced his voice. (Travis was involved in the process.) This video was interesting, but not what I expected.
Hi Fil,
Very interesting look at AI generated music. Since so much music we hear today is using pitch correction the songs sounded like today’s music, what we hear on the radio, we have almost become desensitized to it. But then there were parts that definitely sounded robotic and mechanical. Thanks for doing this one and I guess we will see where AI goes from here. Nice job… Debbie ☮️
I am imagining a Monkees kind of situation (meaning a created band). AI creates some songs, then a band is assembled to copy the AI generated song. I think I just wrote a Black Mirror episode.
There was also the Archies TV cartoon which had real songwriters and studio musicians and singers which had a top radio hit!
The Twilight Zone-like twist at the end will be that the band is a surprise success, but the AI files suit for copyright infringement.
There is a theory that Bach and his "The Well-Tempered Clavier" was just a work of AI computers. It's a pat of "Mudflood" russian conspiracy theory. So if this theory is true, all western music culture was based on AI. And if it's not true, Bach still made first steps to electronic music and AI songs. His "Equal temperament" revolution was unification (some say castration) of real Pythagorean tuning.
stock Aitken and Waterman lol
I agree, the Monkees sprang to mind for me also. It is just another step further down that road of fake bands as a marketable product.
That's one reason I hate remasters. They tend to pitch correct and compress the audio. Then of course streaming services only keep the new versions.
Fil, for machine learning input its called "Prompts" not "Keywords" (Maybe you will think "Whatever dude" haha) As always, great analysis!
Phil Collins once wrote a song about Udio and his ex-girlfriend Susanne who took legal actions against him. I think it was called "Sue sues Udio".
😂
Oh my!😯 Hilarious!!!🤣
.....I thought he reenacted how the title came about on Letterman; when he was nailing and hit his thumb....
or the B-Side "In the Ai'r Tonight"
🤣🤣🤣
This whole idea makes me uncomfortable
Me too
Look up the ai version of hall and oats singing I like big butts and I cannot lie
Creepy to the max.
Sorry if this sounds harsh, but you'll just have to get used to it. The genie is well and truly out of the bottle and isn't going back. Music is just the thin end of a very long wedge. The scope of AI is absolutely enormous: pretty much anything that can be learned can be done.
I'm sure many people had a similar reaction to new impactful technologies, including the printing press, photography, television, etc. Its here, it's going to improve rapidly, and it's going to be the norm.
It probably is not thinking about the A440 standard because it wouldn’t need to. The numeral networks will just adjust the weighting, means, and other distributions of key parameters as it trains on the data. AI Phil coming soon? Hopefully not.
I am imagining a time in the near future when people will hunger for the chance to leave their phone at home and go out to meet real people and listen to real music.
Dream on…
😂
@@haukionkannel nahhh- I think human needs will rise to the surface...
Fil entered the Twilight Zone.
Fil, this was extremely interesting and informative! I must admit that the use of artificial intelligence is very frightening to me ! Having computers that are learning faster than humans is just not the way I think of life ! I could go on forever with this subject so I will stop myself here! I am so happy that you did this video ! I'll be thinking about this and all of the points you made for days now ! What are we going to do when AI learns to feel human emotion ? 💜
great video. Fascinating stuff
So the question is- if you put in exactly the same keywords, and ask it to generate 50 songs from those, will all 50 songs sound the same??
Very interesting Fil
Thanks for this one
Thank you fil. I have learned so much from your videos and enjoy everyone. Love your honesty.
Really great analysis. It's good to see someone taking it seriously and willing to look at the structure without an emotional response. I have made quite a lot of tracks as experiments that were far from derivative. As usual you need to put effort in to get interesting results. Anyway, I'm not concerned. I make music because I enjoy it and my friends all feel the same way. Liked and subscribed.
What stands out to me is the subject. It's about a machine singing.
Hey Fil , how about looking at Alison Kraus's voice? Her singing is so perfect. Maybe one of her duets with Robert Plant would be interesting to compare the two voices and styles. She actually sounds like autotune sometimes, although I know she doesn't use it.
You know she doesn't use it. That's very comforting. Are you her record producer?
@@laveritaforza108 She sounded like that before autotune was invented.
@@joannaneale9816 Invented or when revealed to the masses?
She's out here somewhere!
@@laveritaforza108 What, are you a conspiracy theorist?
In a couple of years we will have chart hits without a human touch - apart from the programmer and the people making money from it. This is going to be killed by copyright lawyers who discover that AI will rarely produce something new but use existing music to train the AI. And existing music has a copyright.
They can't copyright ai songs so it would be hard to sell and make the charts. We could just take it if we like it.
@@TheBagOfHolding this is why the money is in paying for streaming services, not individual songs or albums. The industry has positioned itself for the future monetization of AI media in every way. It is one step down the road of elimination of human made entertainment.
Next step: ask the AI to create the same song in the pre-autopitch era, or a prog rock song, or a more realistic song.
edit: Sorry, you did this for the 2nd song.
It's amazing. Now that people have been trained to enjoy music where human singera have all been auto-tuned/pitch corrected, humans are no longer needed! The computer sounds just like any modern pop singer.
This is the future unless we stand strong and only support real musicians and bands. Live performances are the best. 💪🤘🎸
And if artists demand their vocals not be manipulated.
And if artists demand their voices not be manipulated.
Hi Fil! This whole thing with AI is a bit scary. I hate to think that the future of music will all be "perfected" by AI. So, I personally am against it! Thank you, Fil, for this interesting analysis!
It obviously has a lot of room for improvement, but I think this technology is so cool. It's hard to predict what changes it will bring. Will it take away opportunities from some musicians? Maybe. But it might also provide opportunities for some musicians and non-musicians to express themselves in ways that weren't available to them before.
you are awesome dude
Record companies must be very excited at the prospect of producing and releasing "music" that will cost them very little to produce, and with no royalties to pay!
Worse is the notion that this feeds in perfectly to the listening habits of the young masses now: playlists. Playlists of a type or style of music, rather than being passionate about and following an artist. "Alexa, play some country" etc. It might as well be AI generated if they aren't interested in the band/artist anyway.
Yup. This is largely what the screenwriters' strike was about.
@@TheRealDrJoeyI was going to say the same thing.😄
Yet it wasn't those "young masses" who invented autotune or generative AI music and art...it was people from the older generations and offered it to the younger generation. People aren't born with knowledge, they have to be taught this stuff, and that comes from parents and society.
It's official. Most of the arts (music, film, graphic design, fine arts, photography) can be replicated and it will only get better as time goes on. So the next question is.... What are human artists going to do once they've been replaced? The same is already happening in dozens of industries. What is society's plan for humans when they become unnecessary?
Read Kurt Vonnegut’s “Player Piano.”
We can all become AI copyright infringement litigation lawyers. They'll be necessary. Unless AI takes over that field too.
@@TheOldTapeArchive It will.
@@thos1950 The humans lose.
The only ones who win are the elite. Having a well-stocked bunker on some privately-owned island won't help when ocean levels rise, though :P
You read my mind. Just yesterday, I started wondering what an AI created song would sound like. With pitch correction and every other synthetic thing going on now, it was only a matter of time before this stuff turns up on the charts. Thanks for sharing this.
It's only a matter of time before live bands start covering songs by AI bands. So glad I'm old sometimes....
Back in the eighties I began to notice that they'd replaced the drummer with a drum machine/ computer and the beat remained at the same tempo for the entire song. Of course this is nonsense because a drummer speeds up and slows down with the dynamics/ emotions within the recording of the song. That's what gives it it's feel. No-one batted an eyelid as the programmed drum machine became an industry standard. Should they use a real drummer, he's obliged to play along to a computer giving him a constant beat per minute to emulate. I personally think the end result will be pop stars who don't exist ( holographics ).
I used to go to Toys 'r' Us to buy board games...up until the early 90s, there were dozens and dozens of various games for sale. After that, though, it was whittled down to maybe 15-20 different games...why? It was cheaper to stock fewer selections, so variety suffered. The same thing happened to comic books--there were thousands of titles until publishers wanted to reduce returns on unsold copies, so they stopped accepting them, and that narrowed what publishers would sell to only big titles.
In college (early 90s) I studied right at the cutting edge of digital photography and Photoshop. My professor said it would put film companies like Kodak and darkrooms out of business...and it did. Before that he'd had to go to typesetters unions and tell them that printing was going to change when computers could be plugged into a newspaper-size printer, and he was laughed out of the building. Now there are no typsetters left and computer/printer interfaces are a normal thing.
All in the pursuit of profits and streamlining the "business model" so a small number of people can get richer at the expense of everyone else. It's all like that story about a frog being boiled slowly...
To an audience trained on pitch corrected vocals , they will be receptive to AI music. And the algorithm will deliver personalized music to the individual. And in such a scenario- generic music created by an artist or group will sound - literally- generic.
One could imagine live audiences preferring faultless deliveries - by AI - as opposed to flawed artists
How depressing
Music tuned to achieve desired outcomes in human subjects.
A topic first addressed by the avant garde Canadian film, Strange Brew
Really enjoying your channel and learning a lot.
We will see the first AI superstar very soon. People’s ears are already numb and the ability to market itself and place itself on suggested music/videos will help it. The superstar will be able to do every night show and appearance/concert all at once and even in person if you pay enough. No talent is necessary and that’s been true for many years.
"I'm sorry Dave I'm afraid I can't do that what's the problem I think you know what the problem is just as well as I do what are you talking about this mission is too important for me to allow you to"...
This is exactly what came to mind. What have we done?????
I haven't watched the whole video yet but I want to say that a combination of AI and Streaming will destroy the current income of many writers and musicians. What I hope this leads to is more live music: people on a stage playing instruments and singing. Time will tell.
Like vaudville
@@TheBagOfHolding No, I meant musicians playing music. Gigs.
I can`t handle doing solo stuff anymore because of the ignorant requests...unless I can get 30 to 50 feet away from the drunken mobs.
@@seasideman maybe they can use holograms
@@baneverything5580 People asking you to play songs you don't like or know? I still remember when crowds used to shout for Freebird.
These days there seems to be a lot of emphasis on vocals. A great song for me is Reflections Of My Life (by Marmalade). _So much more_ than just vocals make that song great. When AI can generate _that_ content, I will truly be impressed!
As someone who grew up as a musician and later found a lot of enjoyment in creating midi music, this could turn out to be a really fun way to imagine your own versions of the type of music you love.
ie. Create a song in the style of Zappa using Gentle Giant vocals and intricate King Crimson time signatures... Fun 😊
Great analysis Fil. Hmm well AI is scary. Thank God people have different DNA. 🤣 @wingsofpegasus