Picture this: I'm in my lab, trying to clone my voice like a mad scientist, but then disaster strikes - I catch a cold! As a result, all the recordings I made sound like they were coming from someone with a stuffy nose. But you know what they say - when life gives you lemons, make lemonade! So I decided to give the cloned voice a listen and to my surprise, I actually liked it! It ended up being the voice I used in the video, and honestly, it's kind of grown on me. Who knew a little sniffle could lead to such a happy accident!
The iconic photo at 1:44 was certainly a moment in music history. If you have x-ray vision, you can see a lad in his late 20s in the top center (3rd row). He was apparently unaware that a photo was being taken :) I am that lad! This was my Forest Gump moment. I was at this NAMM show at the invitation of Dave Smith of Sequential Circuits. The reason I was invited was that I had developed a pre-MIDI hardware interface on the Apple II computer. With that hardware I had connected my Apple II to a Prophet 5 synthesizer (with proprietary digital interface). I had also created computer sequencing software to run on the Apple and "talk" to the Prophet 5. That system was on display in the Sequential Circuits booth at the NAMM show.
When i was about 10, I asked my dad (a musician) a question he couldn't really answer, and that was "Will the world run out of music combinations". Bit of a weird question and I still dont know where it came from in me, but I guess it was a case of seeing all of the combinations as having a certain number of variations. What I didnt realise at the time, was that music is not just a sequence - the best music ever made contains emotion, personality, highs/lows, complete changes in tone and delivery and most of all, can actually be a signature. What I see here does blow me away, but sadly brings me back to being 10 and just answers that all gaps wil be closed, but with a lot of ingenuity and emotion lost to boot!! I am not a musician by any means, or a natual talent - Like my late dad, i am a dabbler, love music and am entirely self taught, with no actual abilty to even read music. For me though, that doesn't mean you cant be creative, expressive, make sound, or even glue bits together, to make something new, it is about creativity. I work in IT, have been investigating AI for ages, and looking at applications as well as risks. But for me this would just be a cheat. I dont care if what i create is crap, which is why i dont post any of it anyway! But! If i was to create something crap and use AI to make it good, who would really know if it was you or a bot. What this does, is completely mask individuality, creativity and true emotion of a human by creating an average based on a little experience. The only reality check for true musicians is plaing live, so I feel really sorry for my kids and the future, if using AI to make music sound real and make money, is the only true way left for humans to express one of the oldest ever forms of emotion, bond, story telling, and record of their ancestral history. TBH we are all screwed anyway, but this is just another early nail!!!!!
I don't agree with your last sentence (but then again your name is merchant of doom, so what did I expect 😅) Otherwise some very good points made here!
no just timbre and pitch combinations are infinite....is like asking if you can cut the number 10 in half forever or if you will eventually run out of numbers
It is strange why there aren't more A.I. MIDI generators, tbh. MIDI seems to be the perfect fit for A.I. generation - the entire lexicon of music is well understood and presumably easy enough to input to an A.I. generator. Instead, we get inadequate VST plugins (generators, players, etc) which never really seem to do much of anything useful. What exactly is the problem with MIDI and A.I. generation? Why haven't we seen a premium instrument using the tech?
@Chachi's Musical Experimentation What I see as an ideal compromise would be generating a harmony over a hummed voice, with tweaks possible. Humming a voice and getting a midi output already exists, but AI could refine it, and adding a generative harmony element to that could genuinely really up the simplicity of the process. Generating a melody ain't where it's at, though. Sure, I have no doubt that the tech is gonna get to a point where it won't have any issues doing complex shit on a whim considering the insane growth curves, but like, that process is so uninvolved that it's honestly kinda dull. I love the process of coming up with a melody, automating it makes things boring. It's the same thing as with art. You'll get a nice end product, but no real attachment, since you learned little to nothing about its development. I'd almost think this would be best suited for games, if not for the fact that you usually want something to be musically steady. The coolest thing I could conceive of would be automatic generation of music based on themes for situations, characters, weapons, ect being melded on the spot to resound when in action, but that would almost absolutely consume enough processing power to slow the game while doing that. Also, the music would become kinda predictable.
Because midi only sucks for production. There are no effects! No filters no tone shaping. No samples. No info about what an instrument should sound like. Better use some tracker style files, a lot more promising iMO
For composing music great knowledge is needed. By dispensing knowledge for tools you get something which is not needed at all. Cause it is devoid of imagination... just an unbiased algorhitm.
My electronic music Love started with the SID Chip on my Commodore 64. Loved listening to Sid versions of popular music and Rock hits. Those Sid tunes got me into alot of diffrent bands because I wanted to hear the actual versions of some songs. Then when I got my Windows 98 PC in 1999. My neighbor had hundreds of midis and he gave me a copy of all of them. I still have them and even have some favorites on my phone. Electronic produced music is great and nor to mention all the DOS games that used midi.
@Frizzil No I haven't. I'll have to look that one up. I started messing around with that Basic Pitch website. As long as there are only about 2 instruments those songs convert pretty well. I've never actually edited a Midi before. I'll nerd to find some free software for that
Creative blocks exist because having infinite options. Try using 1 sound 1 octave, 1 finger of human quantized expression. Limitation , constraint and flow. How cool is that you will be blown away. Dope yo.
Hey MAM! Thanks for the insightful comment. I couldn't agree more - constraints can actually fuel creativity, as they force us to think outside the box and come up with unique solutions. I'm super stoked to give this "one sound, one octave, one finger" challenge a whirl. Embracing limitations and focusing on a simpler setup can indeed lead to astonishing results, as we channel our creative energies into a more focused output. I'll definitely give this "minimalist approach" a shot and see where it takes me. Thanks for the dope suggestion! Keep the creativity flowing, my friend. 🎹🚀🧠
It's really nice to see someone highlighting the creative possibilities of this new tech, rather than tearing it down like some futuristic version of analog vs digital.
As a producer, I think it can be good for generating ideas. A.I. can't be very artist and the results are mediocre, but it can definitely help out the process.
@@donbelisario8811 Yes I think it’s achievable. The biggest problem the AI I think would have is being original and tasteful sounds to human ears. This is the same case for drawing AI’s. It can copy, but it can’t make tasteful decisions. The AI would have to become close to humans to do this.
the problem is that a bunch of fucktards are going to use AI to "make music" (I am intentionally using douche quotes") and it will be even worse than it is now - ie the market being saturated with copycats using pirated DAWs and sample packs and following the same generic youtube tutorials. AI and music do not belong together at all
@@Leoppassion Unless an AI has the freedom to develop a sense of taste or preference. As humans over time we develop ideas of what we like and don't like, what is tasteful and what is not to us. Fundamentally an AI doesn't care, but it could be infused with some ground rules of likes or dislikes, and also perhaps more interestingly, be given more abstract rules that aren't likes or dislikes per se but that would consequentially and necessarily lead to the development of them. It would then have what would be considered to be its own style, taste and so on that it would never have been taught and would have been self determined. Also already we have AI's such as the diffusion models that aren't really copying, but that have learnt concepts and know how to get there given text prompts. You could also imagine infusing an AI with a few basic style ideas such as liking curves and circles, negative space, more muted tones than brash colours, caring about it's instagram popularity, being told that when to post and how often matters, consistency is a good idea, and then being given an instagram account with a mission to become as successful as it can. Then coming back in 3 months and seeing how it's getting on. And the initial infusion of ideas, well that it could have sourced itself with the help of AI buddies. Being able to spawn a new AI instance, leave it to its own devices, and coming back to find that it has created a following and life for itself is entirely possible, and maybe even useful because as humans, we might find things that we could learn from whatever communities it has joined and become part of, what content it has produced and settled upon because it found it to be successful.
Fortunately, AI is not yet ready to replace us for that. Yes, it works for 4/4 kick drum snare, a bass and an "identifiable" rhythm section. That is to say 3 tracks maximum including the bass. For the moment, the "AI", which is not an AI here because it only recognizes frequencies = note pitch. 0 emotion. Little nuance and no other distinction for groups of instruments, variations, voices and in other more "complex" musical styles. Finally, the work needed to convert a simple music seems much more time consuming and tasteless than a human transcription (which is much more fun to do).
I don't understand how people NEVER talk about Abundant Music, it has been around for ages and can produce full length songs. It requires some music theory knowledge to set it up to give you results that are specifically what you want, but if set up to your liking the results are always great.
I love the energy in this video! Superlatives and exclamations in a monotone voice works so well! No, I'm not being sarcastic - I really think it works well. Thanks for making this great video!
@@farrael004 Yes, and it's very well done, although I suspect it must have been art by accident discovering that the AI does not take into consideration the emotion of the writing, and so when the writing of this piece is exaggerated in a colorful language, it makes the outcome strangely appealing since the words and the emotion of them are a delightful mismatch 😀 Wonder if they use ElevenLabs?
neither am i... what a load of crap this is , in my mind. wasting your time for getting uninspired, mediocre nonsense patterns, instead of trying to write a song (trying to express your feelings), or learning something bout music,.. but just my 2 cents...
To each their own. Personally I can see it as a tool for components or inspiration. Pure music? Some people are easily pleased. It takes 6 months and 18 minutes per day of playing and practice to be in the top 15% of players. This applies to all habits we practice via rota ie: hand, eye, ear and frontal lobe coordination. If pushing buttons is music and it floats your boat, enjoy. I’m betting at some point you’re gonna want to sing or add some other live or human element to the random of AI. “Oblique Strategies” look it up…
This is a very elegant, probably the most elegant music Ai related video in the history of time so far. Voice over cadence, graphics, information outline, clarity! Very elegant work. Ai in my opinion is just another helping tool for musicians, I do not think it will do more than say LPX drummer or any of the sequencers in the market today. Ai will always complement workflows rather than be an independent music creation tool. Though music making, composing benefits of lots of tech today, still musicians find only meaning in their original ideas and concepts they generate. From my experience, the more Vsts, DAWs, controllers and synths I add, the more lost I am, as if help works backwards with music. The simpler I go the more melodic and creative I get. I’m curious to see what people will make of AI in the upcoming years. Thank you for this very elegant video.
There is of course a very simple explanation for that. The space of tunes is so small it is difficult to generate new ones that the music industry wouldn’t consider copyright infringing, especially if the model is trained on existing music, so the devs thought releasing it would open them up to a bunch of lawsuits. In other words, blame the overly aggressive music industry, especially universal music group.
I dont know man, i predict a Wave of new "Artist and musicians" will arise who wont know what they are doing but are still able to complete Tracks. I guess this will end in a cicle of Quality Regression because why would one learn the stuff in hindsight when the Motivation to do it was never there?
Picture this. AI is trained on already existing material. Once the AI will be widely used by the producers and songwriters, who will create something new? Who will get a new idea in music or songwriting. And even if, why bother when AI will use these ideas immediately, faster than any copycat to generate a song. We will get stuck in a never ending loop of the same ideas. There will be no progress. Another thing is that music and art in general is a way of communication between human beings. A songwriter, producer, musician or even a filmmaker, a painter a writer or any creative person. They all want to express their feelings and ideas to other people through their art. AI will rob the artists from their ways to communicate and express themselves to other people. It's like preventing you to be able to speak. And listeners, they will be robbed from the context of the music it was created and based on. Blues, disco, classical or folk music. Those styles originated from a cultural background and context of their respectable cultures, places and times. Perhaps AI will be able to create something original and new, but without the context, will that have any value?
Great video! I like to use morse code to midi generator for making randomised rhytmic patterns. There are few generators online that work well by inputing text, which then translates to morse code and has few parameters for fine tuning the outcome :)
The reason these tools are likely falling away is because there isn’t yet an answer on who owns that rights to the generated melody, some of the T&Cs of these providers (like OpenAI) either disallow this or possibly imply they do. Who in your opinion were the composers, just yourself or yourself and the AI service? Food for thought
hopefully artists with $$$ do something to protect the industry. this seems cool for people that can't improvise, but honestly the "ai" music sounded awful. i don't think this is going to work because the "ai" will eventually spill its own creations into itself and become a generic oroboros. the same thing will probably happen with visual art. it's already happening with social media bots and generated articles.
This is a great video, thank you for the info and great fun presentation. I have been experimenting with using randomness as a precursor to melody - taking random samples which then become phrases, and then working them into workable melodies. Not exactly AI, but it's a way to get inspired by a phrase you might not have heard before. That's the beauty of something like this - not that it can generate it, because on it's face it's pretty vanilla - but to then take those vanilla ideas it generates and craft something from that, as from a lump of clay to a pot. Until AI learns the best musical ways to manipulate human emotion, AI music will remain flat/lifeless. It is we who breathe the life back into it.
many reasons, but here's some: one, for the artist to then build on top of it, skipping a part of the process they are less interested in doing at this moment. two, art is not always made for the sole sake of making art. unfortunately, in our current economy artists have to sell their work, and AI can help getting results faster. three, a person who has no or just little talent in a particular art. Imagine an amazing programmer who is able to build a really fun game, but is not able to compose any music for it. AI can help offset that lack.
Bro, your video style is fucking awesome! I love the monotonous voice and little quirky exclamations and sayings sprinkled in too 😂 Good shit, I subbed. Also, this was a very enlightening video, thank you good sir
now i want to put that midi file into another AI that will make its own interpretation of what music instruments are being played and create a song out of it. kinf of like color photo to black and white photo and then use an ai to colorise the photo again
There used to be a Midi Maker program that would assign realistic instruments to your midi inputs. I think it was taken offline years ago. It was way ahead of its time. Does anyone know what I'm talking about?
I've been using AI to generate lyrics for my band (I write great rock riffs. I write horrible lyrics). I write in a prompt and a band style, it spits out a pretty forgettable initial pass, but I can take that as a skeleton of the idea and then start revising it. It's really helped me get back into writing my own lyrics again - and helped craft some great lyrical ideas. WOuld I take credit for those lyrics if a grammy was awarded to the song? Okay first of all, hah, it'll never happen, and secondly, of course I would.
No, you should give credit to the thousands of legitimate artists your AI scraped to write your "lyrics". Lame lame lame lame lame. You need to be digging deep inside YOURSELF for your voice. You're only cheating yourself and your audience doing what you're doing. Selling your Soul, you are. You have been warned.
Man I do the same! The choices for titles and themes are endless, and you really have to tweak it to your vision. What’s cool is getting ideas and inspiration you never would have thought of and then inputting those back for more ideas. It’s like the old 4 track days bouncing tracks down to make room for more tracks! I could see a concept album being born from this workflow.
This video is fucking great. It’s actually one of the best and most fun videos about creating with AI tools that I’ve seen, and it’s given me some nice tips (I guess) for making music, or at least for starting messing with it. Well done!
nah, ableton can trnasform that audio to midi natively since like the ice age, its so silly he uses an online tool for that to then improt it into ableton omg
LA Music Composer sounds incredible, I would like to try but I didn't undersand a thing when you explained it. How does it work? Can you do a video with more detail on that please? Many thanks
Creation??? This a grotesque example people being dazzled by the toys instead of by the content of the art they claim to love. It epitomizes brain rot laziness of those who have chosen or brainwashed to be creatively barren. There are countless piano pieces far superior to that piece of crap being played at 7:45. that the narrator claims is "beautiful". Yet, people will ignore all those works in favor of some garbage just because it can be puked out by some a.i music algorithm. There are countless pieces of music that are mind-blowing that were made by geniuses yet people won't listen to it...yet they're throughly impressed by music made by a machine because this machine is "smart"...it doesn't make any sense except that the creatively barren can continue to plunder other peoples work in the most blatant simple minded ways and saying "look what I made aren't I so cleaver and creative" Technology and music have always had a relationship and it can be of immense value (especially to those who may have a disability in playing instruments) but this is disgusting. What a joke.
There is nothing said about copyright and licensing, which might be an important question for some people. If you for example upload a song with digital rights to a website to extract the midi, this might already been a forbidden replication and sharing of a protected arts project. The second problem is about rights resulting from an AI supported project. There were discussions ongoing in the last month, that you do not own the full rights on any AI generated midi and the resulting projects, if i understood the law right. So this might be an easy thing for hobby musicians, but more professional producers, or the ones, that want to become some will have to care about these questions!
As a percussionist myself, I can't see any functional difference that would deem the "humanized" version of the MIDI pattern more realistic, as it still is tied to a grid and lacks things human players will do like time drift, overcompensation for time drifting, as well as velocity and timing changes based on reactive playing to the music.
Everything you mentioned is available in MIDI. You can record yourself playing drums into MIDI and it will record your exact timings and velocities including any time drifts. Many drummers play to a click track when recording. That is simply the grid you are talking about. But you can record someone playing free without a click track or metronome and map it to MIDI as well.
@@High-Tech-Geek My point is not that MIDI can't emulate those things. My point is this AI driven humanizer still spits out drums that sound unrealistic. And that could be the case for several reasons. Either whoever wrote the part doesn't know how to write parts a drummer with two arms and legs is capable of doing, or they are inhumanly impossible to play or the plugin simply gives you a set of fixed offsets that any DAW can do with their own humanizing feature. Time drifting is much more complex than simply shifting the playing by a couple of ticks on either side of a grid. If you want a good example of one of the best humanizer features in a VST drum, get DrumGizmo (it's free and open source), set a beat and play with the humanizer features a bit. And no, that's not AI. That's logic programmed by people which understand how humans play drums.
@@BrunodeSouzaLino If your criticism is of AI, then sure. The examples given here are horrendous. But there's nothing that you've listed that can't be emulated perfectly by MIDI or a better AI. If we're not there yet, it's coming. I'm not sure what grid your referring to. My DAW has timing adjustments down to 1/24th of a 16th note. We're talking 1.3 thousandths of a second granularity at 120bpm. Anyway, right now, I look at AI generated stuff as a tool to start from. I can then inject humanity into it myself. It's a great tool for fresh ideas. But on a different topic, I don't think we should rely on AI to be factual or useful for teaching, as it's just responding via probabilities. It generates so much misinformation and I don't see a fix for that.
If someone would care to open-source an AI model that can MIDI, it would be totally feassible to run and probably even fine-tune on consumer hardware, as seen with stable difussion for images or LLaMa for natural language.
Is this an AI clone of your voice?
Picture this: I'm in my lab, trying to clone my voice like a mad scientist, but then disaster strikes - I catch a cold! As a result, all the recordings I made sound like they were coming from someone with a stuffy nose. But you know what they say - when life gives you lemons, make lemonade! So I decided to give the cloned voice a listen and to my surprise, I actually liked it! It ended up being the voice I used in the video, and honestly, it's kind of grown on me. Who knew a little sniffle could lead to such a happy accident!
@@nobodyandthecomputer That’s pretty cool. Which tool did you use to clone your voice and make it so that it can be used with text to speech?
@@teegees probably Elevenlabs I believe 🤔
@@nobodyandthecomputer that text looks like A.I. too
The voice is horrible
The iconic photo at 1:44 was certainly a moment in music history. If you have x-ray vision, you can see a lad in his late 20s in the top center (3rd row). He was apparently unaware that a photo was being taken :) I am that lad! This was my Forest Gump moment. I was at this NAMM show at the invitation of Dave Smith of Sequential Circuits. The reason I was invited was that I had developed a pre-MIDI hardware interface on the Apple II computer. With that hardware I had connected my Apple II to a Prophet 5 synthesizer (with proprietary digital interface). I had also created computer sequencing software to run on the Apple and "talk" to the Prophet 5. That system was on display in the Sequential Circuits booth at the NAMM show.
First 🥇🏆 to reply here.
:p #90sKid
What do you do now?
Super cool! Thanks for sharing your story here
ohh thats cool
When i was about 10, I asked my dad (a musician) a question he couldn't really answer, and that was "Will the world run out of music combinations". Bit of a weird question and I still dont know where it came from in me, but I guess it was a case of seeing all of the combinations as having a certain number of variations. What I didnt realise at the time, was that music is not just a sequence - the best music ever made contains emotion, personality, highs/lows, complete changes in tone and delivery and most of all, can actually be a signature. What I see here does blow me away, but sadly brings me back to being 10 and just answers that all gaps wil be closed, but with a lot of ingenuity and emotion lost to boot!! I am not a musician by any means, or a natual talent - Like my late dad, i am a dabbler, love music and am entirely self taught, with no actual abilty to even read music. For me though, that doesn't mean you cant be creative, expressive, make sound, or even glue bits together, to make something new, it is about creativity. I work in IT, have been investigating AI for ages, and looking at applications as well as risks. But for me this would just be a cheat. I dont care if what i create is crap, which is why i dont post any of it anyway! But! If i was to create something crap and use AI to make it good, who would really know if it was you or a bot. What this does, is completely mask individuality, creativity and true emotion of a human by creating an average based on a little experience. The only reality check for true musicians is plaing live, so I feel really sorry for my kids and the future, if using AI to make music sound real and make money, is the only true way left for humans to express one of the oldest ever forms of emotion, bond, story telling, and record of their ancestral history. TBH we are all screwed anyway, but this is just another early nail!!!!!
I don't agree with your last sentence (but then again your name is merchant of doom, so what did I expect 😅) Otherwise some very good points made here!
no just timbre and pitch combinations are infinite....is like asking if you can cut the number 10 in half forever or if you will eventually run out of numbers
It is strange why there aren't more A.I. MIDI generators, tbh. MIDI seems to be the perfect fit for A.I. generation - the entire lexicon of music is well understood and presumably easy enough to input to an A.I. generator. Instead, we get inadequate VST plugins (generators, players, etc) which never really seem to do much of anything useful. What exactly is the problem with MIDI and A.I. generation? Why haven't we seen a premium instrument using the tech?
@Chachi's Musical Experimentation What I see as an ideal compromise would be generating a harmony over a hummed voice, with tweaks possible. Humming a voice and getting a midi output already exists, but AI could refine it, and adding a generative harmony element to that could genuinely really up the simplicity of the process.
Generating a melody ain't where it's at, though. Sure, I have no doubt that the tech is gonna get to a point where it won't have any issues doing complex shit on a whim considering the insane growth curves, but like, that process is so uninvolved that it's honestly kinda dull. I love the process of coming up with a melody, automating it makes things boring. It's the same thing as with art. You'll get a nice end product, but no real attachment, since you learned little to nothing about its development.
I'd almost think this would be best suited for games, if not for the fact that you usually want something to be musically steady. The coolest thing I could conceive of would be automatic generation of music based on themes for situations, characters, weapons, ect being melded on the spot to resound when in action, but that would almost absolutely consume enough processing power to slow the game while doing that. Also, the music would become kinda predictable.
because AI is a buzzword marketing gimmick and not actually useful
He's using AI to do tts because it in some places sounds slightly monotonous?
@@kobey. Entirely wrong
Because midi only sucks for production. There are no effects! No filters no tone shaping. No samples. No info about what an instrument should sound like. Better use some tracker style files, a lot more promising iMO
"Are you as blown away as I am?"
"Check out how fire this one is."
"Oh my god have you heard this song?"
/deadpanvoice
ai voice
For anyone looking for the song, the video title is "John Coltrane & Stan Getz - Autumn Leaves/What's New/Moonlight in Vermont (Live 1960)".
For composing music great knowledge is needed. By dispensing knowledge for tools you get something which is not needed at all. Cause it is devoid of imagination... just an unbiased algorhitm.
Agreed, Its not even music any longer
My electronic music Love started with the SID Chip on my Commodore 64. Loved listening to Sid versions of popular music and Rock hits. Those Sid tunes got me into alot of diffrent bands because I wanted to hear the actual versions of some songs. Then when I got my Windows 98 PC in 1999. My neighbor had hundreds of midis and he gave me a copy of all of them. I still have them and even have some favorites on my phone. Electronic produced music is great and nor to mention all the DOS games that used midi.
Same for me but with FF7 MIDIs in the 90’s. Btw I really love the Star Control 2 soundtrack! Did you happen to play it around release?
@Frizzil No I haven't. I'll have to look that one up. I started messing around with that Basic Pitch website. As long as there are only about 2 instruments those songs convert pretty well. I've never actually edited a Midi before. I'll nerd to find some free software for that
You should get a SIDstation asap. I happen to have one to sell. ;-)
@@otisobl What is a SID station. I already have an emulator on my PC.
@@SchardtCinematic No piece of software can sound like a SID.
Creative blocks exist because having infinite options. Try using 1 sound 1 octave, 1 finger of human quantized expression. Limitation , constraint and flow. How cool is that you will be blown away. Dope yo.
that's facts bro
Hey MAM! Thanks for the insightful comment. I couldn't agree more - constraints can actually fuel creativity, as they force us to think outside the box and come up with unique solutions. I'm super stoked to give this "one sound, one octave, one finger" challenge a whirl. Embracing limitations and focusing on a simpler setup can indeed lead to astonishing results, as we channel our creative energies into a more focused output. I'll definitely give this "minimalist approach" a shot and see where it takes me. Thanks for the dope suggestion! Keep the creativity flowing, my friend. 🎹🚀🧠
I was wondering about the same thing but slightly different. I have a hard drive full of unfinished tracks. Maybe AI can help? 🤷🏿♂️
It's really nice to see someone highlighting the creative possibilities of this new tech, rather than tearing it down like some futuristic version of analog vs digital.
The 90's was golden.
The joy of "making" music is supposed to be actually making it.
Indeed but this video debates makes no sense ai can produce music😂
Dude what?! How do you only
Thanks so much for the positive vibes, Kevin! You're the reason we keep making content for this lil' channel that's slowly but surely growing!
@@nobodyandthecomputer bruh r u real
LOL. Right now it 808 subscribers.
1000! 🥳🎉
@@nobodyandthecomputer Congrats! Very nice video, I love your style! :)
As a producer, I think it can be good for generating ideas.
A.I. can't be very artist and the results are mediocre, but it can definitely help out the process.
Have in mind it's just the beginning, I think the layers of complexity in which AI can operate will make them
indiscernible to the human ear
@@donbelisario8811 Yes I think it’s achievable. The biggest problem the AI I think would have is being original and tasteful sounds to human ears. This is the same case for drawing AI’s. It can copy, but it can’t make tasteful decisions. The AI would have to become close to humans to do this.
the problem is that a bunch of fucktards are going to use AI to "make music" (I am intentionally using douche quotes") and it will be even worse than it is now - ie the market being saturated with copycats using pirated DAWs and sample packs and following the same generic youtube tutorials.
AI and music do not belong together at all
@@Leoppassion Unless an AI has the freedom to develop a sense of taste or preference. As humans over time we develop ideas of what we like and don't like, what is tasteful and what is not to us. Fundamentally an AI doesn't care, but it could be infused with some ground rules of likes or dislikes, and also perhaps more interestingly, be given more abstract rules that aren't likes or dislikes per se but that would consequentially and necessarily lead to the development of them. It would then have what would be considered to be its own style, taste and so on that it would never have been taught and would have been self determined. Also already we have AI's such as the diffusion models that aren't really copying, but that have learnt concepts and know how to get there given text prompts. You could also imagine infusing an AI with a few basic style ideas such as liking curves and circles, negative space, more muted tones than brash colours, caring about it's instagram popularity, being told that when to post and how often matters, consistency is a good idea, and then being given an instagram account with a mission to become as successful as it can. Then coming back in 3 months and seeing how it's getting on. And the initial infusion of ideas, well that it could have sourced itself with the help of AI buddies. Being able to spawn a new AI instance, leave it to its own devices, and coming back to find that it has created a following and life for itself is entirely possible, and maybe even useful because as humans, we might find things that we could learn from whatever communities it has joined and become part of, what content it has produced and settled upon because it found it to be successful.
Fortunately, AI is not yet ready to replace us for that. Yes, it works for 4/4 kick drum snare, a bass and an "identifiable" rhythm section. That is to say 3 tracks maximum including the bass. For the moment, the "AI", which is not an AI here because it only recognizes frequencies = note pitch. 0 emotion. Little nuance and no other distinction for groups of instruments, variations, voices and in other more "complex" musical styles. Finally, the work needed to convert a simple music seems much more time consuming and tasteless than a human transcription (which is much more fun to do).
I don't understand how people NEVER talk about Abundant Music, it has been around for ages and can produce full length songs. It requires some music theory knowledge to set it up to give you results that are specifically what you want, but if set up to your liking the results are always great.
First 🥇🏆 to appear here.
:3 #90sKid
@@pawelmiechowiecki7901 Yeah, AI is a pretty vague term and that definitely falls under it
I love the energy in this video! Superlatives and exclamations in a monotone voice works so well! No, I'm not being sarcastic - I really think it works well. Thanks for making this great video!
It's an AI generated voice.
@@farrael004 Yes, and it's very well done, although I suspect it must have been art by accident discovering that the AI does not take into consideration the emotion of the writing, and so when the writing of this piece is exaggerated in a colorful language, it makes the outcome strangely appealing since the words and the emotion of them are a delightful mismatch 😀
Wonder if they use ElevenLabs?
lol, but yes.
Made me think of Elliot from Mr. Robot.
"Are you as blown away as I am?"
No. I am not.
neither am i... what a load of crap this is , in my mind. wasting your time for getting uninspired, mediocre nonsense patterns, instead of trying to write a song (trying to express your feelings), or learning something bout music,.. but just my 2 cents...
To each their own. Personally I can see it as a tool for components or inspiration. Pure music? Some people are easily pleased. It takes 6 months and 18 minutes per day of playing and practice to be in the top 15% of players. This applies to all habits we practice via rota ie: hand, eye, ear and frontal lobe coordination. If pushing buttons is music and it floats your boat, enjoy. I’m betting at some point you’re gonna want to sing or add some other live or human element to the random of AI. “Oblique Strategies” look it up…
It is good to see a lot of work still needs to be done for viable ai gen music.
ikr. Im legit worried for my future as a sound engineer. To be honest though, who's really safe nowadays?...
Until the appealing aspect of good art is clearly defined, we need not fear Ai .
I personally play with my digital audio worktation everyday. Sometimes I play with it while I'm thinking about MIDI.
I liked that fat cat slouching on a music keyboard. The piano piece at the end was lovely. Thanks for sharing.
This is a very elegant, probably the most elegant music Ai related video in the history of time so far. Voice over cadence, graphics, information outline, clarity! Very elegant work.
Ai in my opinion is just another helping tool for musicians, I do not think it will do more than say LPX drummer or any of the sequencers in the market today. Ai will always complement workflows rather than be an independent music creation tool. Though music making, composing benefits of lots of tech today, still musicians find only meaning in their original ideas and concepts they generate. From my experience, the more Vsts, DAWs, controllers and synths I add, the more lost I am, as if help works backwards with music. The simpler I go the more melodic and creative I get. I’m curious to see what people will make of AI in the upcoming years.
Thank you for this very elegant video.
Some years later the 90s arrived. Since you were talking about 1982, I'd have a guess at 8 years
Your channel feels like one with a million subscribers, because it's damn good.
This is sarcasm, we love it. I don`t think many realise its a piss take.
no
There is of course a very simple explanation for that. The space of tunes is so small it is difficult to generate new ones that the music industry wouldn’t consider copyright infringing, especially if the model is trained on existing music, so the devs thought releasing it would open them up to a bunch of lawsuits. In other words, blame the overly aggressive music industry, especially universal music group.
we need this open-source or crowdfunded and then it can live on a server in the Bahamas or something, just like sci-hub and genlibru
Did this just become my favourite YT thing? I think so
I dont know man, i predict a Wave of new "Artist and musicians" will arise who wont know what they are doing but are still able to complete Tracks. I guess this will end in a cicle of Quality Regression because why would one learn the stuff in hindsight when the Motivation to do it was never there?
mr. Nobody: You drive final frontier. I hope your mind will be reinkarnated by forward entities.
Picture this. AI is trained on already existing material. Once the AI will be widely used by the producers and songwriters, who will create something new? Who will get a new idea in music or songwriting. And even if, why bother when AI will use these ideas immediately, faster than any copycat to generate a song. We will get stuck in a never ending loop of the same ideas. There will be no progress. Another thing is that music and art in general is a way of communication between human beings. A songwriter, producer, musician or even a filmmaker, a painter a writer or any creative person. They all want to express their feelings and ideas to other people through their art. AI will rob the artists from their ways to communicate and express themselves to other people. It's like preventing you to be able to speak. And listeners, they will be robbed from the context of the music it was created and based on. Blues, disco, classical or folk music. Those styles originated from a cultural background and context of their respectable cultures, places and times. Perhaps AI will be able to create something original and new, but without the context, will that have any value?
Great video! I like to use morse code to midi generator for making randomised rhytmic patterns. There are few generators online that work well by inputing text, which then translates to morse code and has few parameters for fine tuning the outcome :)
One simple impartial question that everyone has to answer themselves:
>> Is it still art ?
I just found you. It's like the video is good, then whoever is narriating's tone and delivery is just next level icing on the cake.
The reason these tools are likely falling away is because there isn’t yet an answer on who owns that rights to the generated melody, some of the T&Cs of these providers (like OpenAI) either disallow this or possibly imply they do.
Who in your opinion were the composers, just yourself or yourself and the AI service? Food for thought
hopefully artists with $$$ do something to protect the industry. this seems cool for people that can't improvise, but honestly the "ai" music sounded awful. i don't think this is going to work because the "ai" will eventually spill its own creations into itself and become a generic oroboros. the same thing will probably happen with visual art. it's already happening with social media bots and generated articles.
AI is good for helping music production, but outright replacing production with AI is practically impossible.
@@struckrex Nothing is impossible.....aaaaah something is impossible but it isn't AI production
This video has mega meta uncanny valley vibes... awkwardly written and spoken, like it made by AI...
This is a great video, thank you for the info and great fun presentation. I have been experimenting with using randomness as a precursor to melody - taking random samples which then become phrases, and then working them into workable melodies. Not exactly AI, but it's a way to get inspired by a phrase you might not have heard before. That's the beauty of something like this - not that it can generate it, because on it's face it's pretty vanilla - but to then take those vanilla ideas it generates and craft something from that, as from a lump of clay to a pot. Until AI learns the best musical ways to manipulate human emotion, AI music will remain flat/lifeless. It is we who breathe the life back into it.
why would you want to automate an artistic process like this, whats the point then
many reasons, but here's some:
one, for the artist to then build on top of it, skipping a part of the process they are less interested in doing at this moment.
two, art is not always made for the sole sake of making art. unfortunately, in our current economy artists have to sell their work, and AI can help getting results faster.
three, a person who has no or just little talent in a particular art. Imagine an amazing programmer who is able to build a really fun game, but is not able to compose any music for it. AI can help offset that lack.
MIDIs. Englishs. Spanishs. DOSs.
😂😂😂
What a time to be alive!
When satire becomes the gospel.
Pure poetry!
Worktation lol. Your voice clone was Almost perfect. This video is excellent, thanks for all the knowledge dropping.
Bro, your video style is fucking awesome! I love the monotonous voice and little quirky exclamations and sayings sprinkled in too 😂 Good shit, I subbed. Also, this was a very enlightening video, thank you good sir
"hey nobodies" your channel has litterally 6k subs bro
now i want to put that midi file into another AI that will make its own interpretation of what music instruments are being played and create a song out of it. kinf of like color photo to black and white photo and then use an ai to colorise the photo again
There used to be a Midi Maker program that would assign realistic instruments to your midi inputs. I think it was taken offline years ago. It was way ahead of its time.
Does anyone know what I'm talking about?
This is some quality work.
Hey, thanks for dropping a comment!
For starters thank you for the knowledge
I've been using AI to generate lyrics for my band (I write great rock riffs. I write horrible lyrics). I write in a prompt and a band style, it spits out a pretty forgettable initial pass, but I can take that as a skeleton of the idea and then start revising it. It's really helped me get back into writing my own lyrics again - and helped craft some great lyrical ideas. WOuld I take credit for those lyrics if a grammy was awarded to the song? Okay first of all, hah, it'll never happen, and secondly, of course I would.
nice job admitting to basically cheating when it comes to creating art. how do you like that ai dick
No, you should give credit to the thousands of legitimate artists your AI scraped to write your "lyrics". Lame lame lame lame lame. You need to be digging deep inside YOURSELF for your voice. You're only cheating yourself and your audience doing what you're doing. Selling your Soul, you are. You have been warned.
Man I do the same! The choices for titles and themes are endless, and you really have to tweak it to your vision. What’s cool is getting ideas and inspiration you never would have thought of and then inputting those back for more ideas. It’s like the old 4 track days bouncing tracks down to make room for more tracks! I could see a concept album being born from this workflow.
What AI do u use for lyrics?
You just got yourself a Subscriber!
This video is fucking great. It’s actually one of the best and most fun videos about creating with AI tools that I’ve seen, and it’s given me some nice tips (I guess) for making music, or at least for starting messing with it. Well done!
nah, ableton can trnasform that audio to midi natively since like the ice age, its so silly he uses an online tool for that to then improt it into ableton omg
This video is like someone read my mind. Thanks for educating people on this stuff
THE WONDERS OF THIRD ORDER PROBABILITY ANALYSIS
Your voice is so pleasant.
And the content of this video was spectacular. Immediately subscribed!
Good. Now make it easy for my great grandma Wizard to do-
Thanks to David Smith for MIDI.
Man... I'm so subscribing to your channel, this is gooooooooooooooold info man, gold! Thank you so much!
man, you should make videos about how to use AI tools to makes movies and videos. ❤
epic squirrel monkey reference in the intro
havent seen them for ages... i wonder what they're doing now?
his constant pluralization of the word "MIDI" really fucked my life up for a second, couldn't handle that shit
TRBLs, MKRs, MIDIs, GIFs, FAQs, LEDs, URLs, PINs, CPUs, PDFs, HTTPs, ISPs, VPNs, DVDs, ATMs, OEMs, SSDs, NFTs
You hit this all spot on. Love it! Gotta' give it a go. Thanks!
midi sampling! Things are changing
WHAT'S THE POINT OF AUTOMATING ART
Thanks, you taught me some things I didn't know that will come in handy. Sunscribed.
most underated chanel
LA Music Composer sounds incredible, I would like to try but I didn't undersand a thing when you explained it. How does it work? Can you do a video with more detail on that please? Many thanks
4:00 Also… Muse-Net not Muse-N-Net.
All your content is really great! thank you!
Awesome video! It's amazing what tools we have at our disposal nowadays for music creation.
creation?
This takes the you out of music creation.
Creation??? This a grotesque example people being dazzled by the toys instead of by the content of the art they claim to love. It epitomizes brain rot laziness of those who have chosen or brainwashed to be creatively barren. There are countless piano pieces far superior to that piece of crap being played at 7:45. that the narrator claims is "beautiful". Yet, people will ignore all those works in favor of some garbage just because it can be puked out by some a.i music algorithm. There are countless pieces of music that are mind-blowing that were made by geniuses yet people won't listen to it...yet they're throughly impressed by music made by a machine because this machine is "smart"...it doesn't make any sense except that the creatively barren can continue to plunder other peoples work in the most blatant simple minded ways and saying "look what I made aren't I so cleaver and creative" Technology and music have always had a relationship and it can be of immense value (especially to those who may have a disability in playing instruments) but this is disgusting. What a joke.
Such a soothing voice
Out painting for midi … wowzaz… just need this for audio next
I see what you did there with "This MIDI does not exist"
There is nothing said about copyright and licensing, which might be an important question for some people. If you for example upload a song with digital rights to a website to extract the midi, this might already been a forbidden replication and sharing of a protected arts project. The second problem is about rights resulting from an AI supported project. There were discussions ongoing in the last month, that you do not own the full rights on any AI generated midi and the resulting projects, if i understood the law right. So this might be an easy thing for hobby musicians, but more professional producers, or the ones, that want to become some will have to care about these questions!
Love the content, smooth lines and thank you! Subbed.
Are you the “wowzers” guy from the Ableton tutorial videos? Or just an A.I. trying to impersonate him?
As a percussionist myself, I can't see any functional difference that would deem the "humanized" version of the MIDI pattern more realistic, as it still is tied to a grid and lacks things human players will do like time drift, overcompensation for time drifting, as well as velocity and timing changes based on reactive playing to the music.
Everything you mentioned is available in MIDI. You can record yourself playing drums into MIDI and it will record your exact timings and velocities including any time drifts. Many drummers play to a click track when recording. That is simply the grid you are talking about. But you can record someone playing free without a click track or metronome and map it to MIDI as well.
@@High-Tech-Geek My point is not that MIDI can't emulate those things. My point is this AI driven humanizer still spits out drums that sound unrealistic. And that could be the case for several reasons. Either whoever wrote the part doesn't know how to write parts a drummer with two arms and legs is capable of doing, or they are inhumanly impossible to play or the plugin simply gives you a set of fixed offsets that any DAW can do with their own humanizing feature. Time drifting is much more complex than simply shifting the playing by a couple of ticks on either side of a grid. If you want a good example of one of the best humanizer features in a VST drum, get DrumGizmo (it's free and open source), set a beat and play with the humanizer features a bit. And no, that's not AI. That's logic programmed by people which understand how humans play drums.
@@BrunodeSouzaLino If your criticism is of AI, then sure. The examples given here are horrendous. But there's nothing that you've listed that can't be emulated perfectly by MIDI or a better AI. If we're not there yet, it's coming. I'm not sure what grid your referring to. My DAW has timing adjustments down to 1/24th of a 16th note. We're talking 1.3 thousandths of a second granularity at 120bpm.
Anyway, right now, I look at AI generated stuff as a tool to start from. I can then inject humanity into it myself. It's a great tool for fresh ideas.
But on a different topic, I don't think we should rely on AI to be factual or useful for teaching, as it's just responding via probabilities. It generates so much misinformation and I don't see a fix for that.
Just "converting" that music to MIDI hurt my soul.
this video is such a gift - liked and subbed
Thanks a lot! Bests from Brazil!!!!
Amazing info dude. Thanks. How can i download the MIDI file generated in the Los Angeles Music Composer ?
Yes I am trying to do the same thing and it only lets me download an 'audio' file.
very intresting toipc keeep them coming...
Soon there be nothing left for any human to do, i think imperfection will be a growing trend. But great tools i will be using thanks for the video 🔥🔥🔥
it took me 7 minutes to realize the narration was AI.
your sarcastic tone is funny haha
Moral of the story is Ai has always been around it’s just evolved!!!
Great video! Thanks for making this!
This is death of electronic music… Good
If someone would care to open-source an AI model that can MIDI, it would be totally feassible to run and probably even fine-tune on consumer hardware, as seen with stable difussion for images or LLaMa for natural language.
i actually made something like this (openmusenet-2) which is an early version of the model in my program
Or you can easily use ableton built-in audio convertor to get harmony, melody or even drum out of the sound.
🔴I love so much the visual style of your midi story, simply boiled down to the essence🔴
I downloaded the magenta plug in and I can't seem to figure out how to input a file for it to process. Anyone else have this problem?
As Grogu said while driving the robot “no no no no no no”
Beautiful rhodes sound? Bahaha. That was ghastly.
Dude! You have no idea how perfect this video is! A thousand thank yous!
hey thanks
Thanks for this video
my current dream since 2 weeks now: Logic Pro AI 🤯
Amazing! Thanks for sharing!
U mad bro
Appreciate sharing the knowledge 😊😊
Aiva is good for generating AI midi, lots of tools too
This was great!
Nice vid bruh, Salute.
Grazie amico mio