The DOMi and JD Beck one was actually pretty on point. Not as in “their finest moment” but certainly something they could have played during a 3-second uninspired stretch on a random evening.
Dom and jd beck mist often in a formula. So you didnt need ai to recreate that. As far as zappa style is concerned. I could randomly picked chords and added a simple drum beat manually got as good of result.
Truly one of the more productive uses of large quantities of electricity I've ever seen. It's honestly hard to see how we ever used to manage using controlled parametric inputs instead of prompts.
The amount of effort that went into this video will have me hard for the next three days. Bruh, do you need a hug dude? I reeaaalllyyy want to give you one. Haha, jokes aside, amazing, AMAZING video. I've subbed and will be all over your content from now on. I'm in love with you. 🖤
Nice - your GPT midi experiments are getting better with each video. Our app offers text-to-midi as a VST and we're actively building toward AI MIDI API integration. It's a long and expensive process so it's taken us a few years but we're getting there. Would love to collab with you some time!
With your help, I've been able to use a midi sequence file and use it as an output on a analog hardware instrument (microKORG). I mean, If I had been told 4 months ago, that an artificial intelligence would be able to generate a sequence capable of interacting in the material world trough an analog instrument. I wouldn't have believed it. THANK YOU SO MUCH.
It feels like the internet is starting all over again. There are obvious dangers, but opportunities for those whom are clever. I believe inspired visionaries utilizing AI will quickly spawn a new set of moguls. Brave new world.
I'm of the Mind music is a reflection of the human experience born from organic elements. Believing that AI will somehow capture the complexity that lies within the human brain that is expressed through word and music is merely a fantasy.
OK my friend, please can you point me in the direction of how you created the voices used in this video, like the one in the intro. Thanks for all your hard work!
Once it can accept complicated and specific prompts with specific constraints. Then I guess ill just bow down to our robot overlords and I'll play whatever they come up with
Text to Chords (only for patreons): I need assistance in producing AI-generated text that I convert to music using MIDI files. Initially, I'll provide a description of the format I need for the textual representation of the music. Since music is a time-based art form, the notes follow each other in time, and sometimes there are no notes, that is, silences. Here I need you to generate Chords, so, more than one note each time, could be 2, 3, 4, etc. The way I would like you to generate them is as follows: Each chord is represented by these elements: The pitches of the notes (integer values). Because I will use this text representation and convert it to MIDI, the note should be a number from 21 (that is note A0-27.50 Hz) to 96 (that is C7-2093 Hz), so use these numbers to represent the note. The duration of the note (float value) represented as: 0.125 for an eighth note 0.25 for a quarter note 0.5 for a half note 1 for a whole note 2 for a double whole note But it could be any number between 0 and 2, because you know, musicians are creative, so why not 0.29 or 1.22, etc. With this format, I need you to generate a text that I will convert into music in this format: chords_pitch_duration_data = [ ((note, note, note), duration), ((note, note), duration), (note, duration), etc ] And when there is a silence, the note should be 0 and the duration is how long that silence lasts. The key is that harmony is the vertical aspect of music, or the combination of different pitches sounding at the same time. Chords are the building blocks of harmony, and they are made up of two or more notes that are played together. The simplest type of chord is a two-note chord, also known as a diatonic chord. Diatonic chords are made up of two notes that are next to each other on the musical scale. For example, a C major chord is made up of the notes C and E. Three-note chords are also known as triads. Triads are made up of a root note, a third, and a fifth. The root note is the lowest note in the chord, the third is the note that is two semitones above the root, and the fifth is the note that is seven semitones above the root. For example, a C major triad is made up of the notes C, E, and G. Four-note chords are also known as seventh chords. Seventh chords are made up of a root note, a third, a fifth, and a seventh. The seventh note can be either major or minor. For example, a C major seventh chord is made up of the notes C, E, G, and B. And you also have: Sus chords: Sus chords are made up of a root note, a third, and a fifth, but the third is replaced with a suspended fourth or second. For example, a Csus4 chord is made up of the notes C, F, and G. Add chords: Add chords are made up of a root note, a third, a fifth, and an additional note. For example, a Cadd9 chord is made up of the notes C, E, G, and D. Augmented chords: Augmented chords are made up of a root note, a major third, and a perfect fifth. For example, a C augmented chord is made up of the notes C, E#, and G#. Diminished chords: Diminished chords are made up of a root note, a minor third, and a diminished fifth. For example, a C diminished chord is made up of the notes C, Eb, and Gb. Harmony is an important part of music, and it can be used to create a variety of different moods and emotions. For example, major chords are often used to create a sense of happiness or joy, while minor chords are often used to create a sense of sadness or melancholy. Please note that AI-generated music may not sound pleasing as it is randomly generated. So, we will use music theory, not random math. Take into account musical concepts like scales, modes, etc. Now that you have a full understanding of the text representation, we will create some awesome music! Are you ready to start generating music? If so, respond with 'YES' and nothing else.
Artificial intelligence still lacks the magic that really moves the spirit. There will probably never be a way to create music the way Ryo Kawasaki does.
As a music producer, composer and jazz pianist, i considered every one of these a certified masterpiece. We finally have access to the best music without limitations.
Made it about half way thru... There's clearly enormous potential, but it's still a long way off putting Ed Sheeran outta business. I'll come back and check out your A.I. MIDI adventures in 6 months.
Whaooo! you´re so good and so fun or was it you´re så (sorry that was swedish from the pure forest mountain in Stockholm where I´m living: GPR that...) so good and so fun. How would I know which is right? Good and fun or fun and good. Not to mention: creative to begin with... Thank you!
The video was great , a real art piece. The generated music was absolutely terrible , but still made for an entertaining vid with all that work put in.
Well, i persevered to the end. Music has been a significant part of my life, from rock through to deepest classical, so I'm accepting of a wide range of the world of music. Despite the hard work you put in, everything here was absolutely horrendous. None, yes, NONE, of the generated "melodies" were even remotely melodic - for any music genre, and the harmonies were similarly utterly useless. Your work on the visuals was great. Take that as a pointer to where your talents reside. But thank you for your efforts. Of course,the vid is around 12 months old, and advances in AI music have been great in that short time, so AIVI may have improved, and it perhaps can also give better results in more musical hands. [And, text-to-music is a demanding concept in itself.]
Still sounds rubbish, though. Call back when A.I. can write something as evocative and melodious as, say, Ravel's 'Piano Concerto in G Major, M. 83 - II. Adagio assai' (with full string orchestra).
I love how you used so many different ai llm tools to generate this video from a optimistic techy standpoint, but on the otherhand this video fucking hurt my brain because jesus christ the outputs were so fucking mid.
The reason why I will never be impressed with A.I. is because it's still just a computer functioning on 1s and 0s. Even though the data base is huge and continues to grow, it's still just a glorified calculator at the end of the day, capable of computing via language instead of numbers.
"Wow, this AI-Powered MIDI Generation in Advanced Mode is mind-blowing! The level of creativity and musicality it brings is simply remarkable. It's incredible to witness how AI has evolved to compose music that rivals human composers. The melodies, harmonies, and rhythms produced are truly mesmerizing. This technology has the potential to revolutionize the music industry and open up endless possibilities for musicians and artists. Kudos to the developers for pushing the boundaries of AI and music. I can't wait to see what the future holds for AI in music composition!" Positive coment made with IA
The prompt: I need assistance in producing AI-generated text that I convert to music using MIDI files. Initially, I'll provide a description of the format I need for the textual representation of the music. Since music is a time-based art form, the notes follow each other in time, and sometimes there are no notes, that is, silences. The way I would like you to generate them is as follows: Each note is represented as a tuple of two elements: The pitch of the note (integer value). Because I will use this text representation and convert to MIDI the note should be a number from 21 (that is note A0 - 27,50 Hz) to 96 (that is C7 - 2093 hz) so use these numbers to represent the note. The duration of the note (float value) represented as: 0.125 for an eighth note 0.25 for a quarter note 0.5 for a half note 1 for a whole note 2 for a double whole note But could be any number between 0 and 2, bocouse you know, musician are creative so why not 0.29 or 1.22, etc. With this format i need you generate a text that i will covert in music in this format: melody_pitch_duration_data = [ (note, duration), (note, duration), (note, duration), etc, ] And when there is a silence the note should be 0 and the duration is how long is that silence. A melody is a linear sequence of notes that the listener hears as a single entity. It is the foreground to the backing elements and is a combination of pitch and rhythm. Sequences of notes that comprise melody are musically satisfying and are often the most memorable part of a song. There are many ways to describe a melody. Here are a few: ● Pitch: The pitch of a melody is the relative highness or lowness of the notes. Melodies can be high, low, or somewhere in between. ● Rhythm: The rhythm of a melody is the pattern of long and short notes. Melodies can have a slow, steady rhythm, a fast, syncopated rhythm, or something in between. ● Intervals: Intervals are the distance between notes. Melodies can use a variety of intervals, from small steps to large leaps. ● Contour: The contour of a melody is the overall shape of the melody. Melodies can be ascending, descending, or something in between. ● Tonal center: The tonal center of a melody is the note that the melody feels like it is centered around. Melodies can have a strong tonal center, a weak tonal center, or no tonal center at all. When describing a melody, it is important to consider all of these factors. The pitch, rhythm, intervals, contour, and tonal center all contribute to the overall sound of the melody. Here are some examples of how to describe melodies: ● The melody of "Happy Birthday" is simple and repetitive, with a clear tonal center. ● The melody of "Yesterday" by The Beatles is more complex, with a variety of intervals and a changing tonal center. ● The melody of "Bohemian Rhapsody" by Queen is highly dramatic, with a wide range of pitches and rhythms. Quality melodies typically limit their range to about an octave-and-a-half, feature repeating elements like melodic intervals and rhythmic patterns, and consist of stepwise motion with occasional energetic leaps. Good melodies also interact meaningfully with the bass line, employing a mix of parallel, similar, oblique, or contrary motions for a dynamic, counter melodic effect. Finally, a standout melody tends to have a climactic moment, often a high note with significant harmonization and strong rhythmic placement, which then descends to a restful cadence. No matter how it is described, a melody is one of the most important elements of music. It is what gives a song its identity and makes it memorable. Please note that AI-generated music may not sound pleasing as it is randomly generated so we will use music theory but not random math so don't randomize the generation process. take into account musical concepts like scales, modes, etc. Now that you have a full understanding of the text representation, we will create some awesome music! Are you ready to start generating music? If so, respond with ‘YES’ and nothing else.
Wait but how did he generate the voices so well.. like in the D Angelo example. Can anyone help me with that part? Clearly that has nothing to do with the midi chords.
I am much more interested in how you got Scarlett Johansson's voice to read the text and how you got a singer to do a whole bunch of complex harmonies in the middle of the video. Where is that coming from? What apps are you using?
That female voice is of Scarlett Johansson. She was the actress in the movie "Lucy" (2014). About a woman who became 1,000,000 times more advanced than a quantum computer. Good choice.
So, I wonder if there would be anything like the kind of prompt mashing that stable diffusion and other image generators have, where you can just throw it a whole heap of names and tell it to make a collab between a wide range of artists to get a mashup style in quite a specific way.
love your vids! I can tell you used musiclm for the music at :25, it gave me an extremely similar sounding piece of music that I then used as inspiration for a piece of my own so its basically embedded into my ear. keep killing it bro
Yes of course LOL. But once you have found a prompt that gets back better musical results, that work is done - save it and you’ll never have to make it again. If someone’s already found that prompt for you, that’s a whole lot of work you’ve skipped already. And, in the future, if somebody makes an online interface that takes your text as input, prompts ChatGPT for you, and sends the created song back to you as a MIDI file, you can now run through this whole process in a fraction of the effort and time. It’s far from being a replacement for human creativity, and it should remain that way. But at this point it is a semi-competent musical idea generator, with some actual interesting melodies/progressions among the noise it spits out. With enough time and better models, the idea generation could become more consistent and the process more streamlined. And I mean, that sounds pretty neat. I know I really enjoyed messing with MuseNet when it was online, so the idea that a general-purpose LLM could someday serve the same role is pretty wild to me.
In this comment section, we have two comments from music producers. One saying trash. The other saying legit on what was spit out. So it’s not about AI, human or nature. It’s how it settles in our ears. Carry on with whatever you are using. Someone will like/hate it. The human story.
Nine months later and all I have to do is type a sentence or two in an app called Suno and I get a fully arranged song with lyrics and vocals. Nine months from now, it will be broadcast quality, separate tracks, and we''ll have the ability to dictate the chord progressions, melody, arrangement, playing styles and singer's voice and style. Nine months after that, we're all dead because all these companies are releasing AI code as open source. Won't take long before the wrong people get it and everything starts falling apart due to crime, fraud, disinformation, political and social unrest and other nefarious uses. Open source means everyone, not just good people. BUT... In the meantime, I've loving what it's doing with music!!
Iget an error with the chords TypeError Traceback (most recent call last) in () 17 current_time = 0 18 for chord_pitches, duration in chords_pitch_duration_data: ---> 19 for pitch in chord_pitches: 20 midi_file.addNote(track=0, channel=channel, pitch=pitch, time=current_time, duration=duration, volume=volume) 21 current_time += duration TypeError: 'int' object is not iterable
ChatGPT was never trained to generate music so the fact that it does anything at all is pretty remarkable. With the advent of systems specifically trained for generating music, it's clear that there's no need for sarcasm anymore, and ample room for distress among the music lovers and professionals...
I tried having GPT generate a number of clips in a Phrygian key. It did… okay after awhile but nothing inspiring. Eventually it gave up and told me I need to try a different algorithm.
You folks did one HELLUVALOTTAWOIK!!! Zappa's piece was pretty close to bein' sumpin'...the rest (sept4 Domi-Beck) was pretty un-inspiring. Let's hope AI doesn't get the creative love & passion that the future artists of the World put their minds to. You really saved us a LOT of time...For that, I AM TRULY grateful!!!
ChatBots are not the right tool. AT ALL. Look, Apple already has something going on this front with its automatic musical pattern generators, even in the humble GarageBand. What’s needed is progress in developing an AI specifically designed for MIDI in a DAW, trained in a variety of music styles, and capable of being further trained by analyzing a piece of MIDI music inside a DAW. So, for example, I should be able to write two or three cues for a TV show, then have the AI tool study my three cues and then generate more cues with the same template in my style. That’s coming. It’s probably already sitting in a lab somewhere in Silicon Valley or Boston or the Rhine or Israel or who knows where.
Bard more than likely slowed down its melodies because it doesn't understand when it's looping a sequence If the next sequence doesn't have the exact same amount of ticks or if there was a duration of time where no note was played in the melody where Bard didn't count the empty space of ticks, will caused it to screw up its timing. Thus if you tell it to strictly stay with in the confines of 60 ticks per 32nd note or 1960 ticks per bar and make sure it counts any empty space as well. It should work. Or simply changing all of the 0.5 and 1.0 to 0.25 would have worked also.
It will sooner then you think. It needs a clear "understanding" of what makes us like or dislike in music, what is harmony, itc. In other words it needs to learn and it will learn much much faster. But stop, maybe we should not allow it becoming superior and just kill it while its not too late?
Well the music is far from being impressive. But I think with a more specialised corpus, better embeddings and prompt engineering considering other dimensions (dynamics, timbre, range, music forms, motive, transition/ponctuation...), it will lead to more exploitable material. It's just the beginning anyway. It's already transforming sound engineering, it will sound transform actual music production. I hope humanity will still enjoy create and perform music anyway... .
Toca como alguien muy experimentado, el problema es que no a toda la gente de la generación actual le gusta la música complex, la mayoría de estos casos es música para músicos, porque si le pones una 7ma de ebmaj, y empiezas de allí, hay mucha gente que nos sorprenden los acordes aumentados y otros que no, a lo que voy es que si te ayuda la ia no te dará el super trabajo 2023 que revolucionaras la industria, solo sacaras música como si fueses un músico experimentado que siento que es el peor engaño que te puedes hacer a ti mismo. Si algun día tocas en un concierto no creo que pongas el pc con la ia para que te toque el pianito, una cosa es saber utilizarlo bien y otro que te haga todo el trabajo que en el fondo te quita merito propio
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
Turn off your computers, take your instruments and make noise. Don't play video games, play BbMaj7. Good God, Real Music By Real Musicians never die. 💜☔️
you have so many cliche, over-used words and phrases that after 2 minutes of listening to you, I couldn't take it any more and disconnected. Consider backing off these worn out spoken words and using simple instead. More message, less ornamental BS
as a music producer, composer, and jazz pianist nearly all of melodies were like literally terrible. sure, they sound decent in some places but there's no structure and none of that human "feel" that actually makes music be a reasonable display of emotion. looks like a fun tool to mess with but in no way ready for production ready music
@@finlay1702it’ll never be able to come close to a talented human. Personally, when I listen to music I always think wow there’s a human behind this what a creative dream they’ve let me into but obviously if you use ai it won’t have any of that
Many actual musically talented humans have difficulty using music as an expressive medium, how well do you think something without feelings will be able to do it?
The massive problem with chat gpt or any other ai is the welfare sized token alottment that you get. Even with gpt4 its only 18,000 tokens . Once your code gets to a certain size the AI simply gets chronic amnesia and your progress comes to a sceeching halt as you run around in circles trying to solve problems while the ai continuosly loses its memory with everytime you post the new scrypt. Claude AI boasts a nice 100,00 tokens but its coding skills and unwillingness to complete anything makes it pretty much useless.
The DOMi and JD Beck one was actually pretty on point. Not as in “their finest moment” but certainly something they could have played during a 3-second uninspired stretch on a random evening.
the only thing i would add is more hihat
Dom and jd beck mist often in a formula. So you didnt need ai to recreate that.
As far as zappa style is concerned.
I could randomly picked chords and added a simple drum beat manually got as good of result.
quite impressed about how much chatgpt knows about them
Beautiful video editing effort and illustrative too. Amazing work!
Truly one of the more productive uses of large quantities of electricity I've ever seen. It's honestly hard to see how we ever used to manage using controlled parametric inputs instead of prompts.
Woah, that's impressive, expecially the chords in the neo-soul style (8:49). It's almost unbelievable that AI composed all of this!
The amount of effort that went into this video will have me hard for the next three days. Bruh, do you need a hug dude? I reeaaalllyyy want to give you one. Haha, jokes aside, amazing, AMAZING video. I've subbed and will be all over your content from now on. I'm in love with you. 🖤
Nice - your GPT midi experiments are getting better with each video. Our app offers text-to-midi as a VST and we're actively building toward AI MIDI API integration. It's a long and expensive process so it's taken us a few years but we're getting there. Would love to collab with you some time!
text-to-midi AND INTELLIGENT MUSIC PROCESSING WILL BE MY PhD Topic, Seems like its going to be fun
Whats the name of that VST?
With your help, I've been able to use a midi sequence file and use it as an output on a analog hardware instrument (microKORG). I mean, If I had been told 4 months ago, that an artificial intelligence would be able to generate a sequence capable of interacting in the material world trough an analog instrument. I wouldn't have believed it. THANK YOU SO MUCH.
It feels like the internet is starting all over again. There are obvious dangers, but opportunities for those whom are clever. I believe inspired visionaries utilizing AI will quickly spawn a new set of moguls. Brave new world.
Amazing Work, Sir! Love you're super unimpressed delivery :)
The editing on this video is incredible. It reminds me of the "How to turn s sphere inside out" video that went viral on TH-cam years ago
Love the combination of exploration and comedy in your videos!
I'd love to see you generate Chopin-style music, I think his music is very nice and it'd be interesting to see how AI does it 😊
You successfully make this video looks like uploaded 30 years ago, but I love it🤣
The AI generated D’Angelo just gave me a shiver(compliment)
9:40 Frank Zappa my fav musician, and one of the greatest minds in humanity ..... wish he was alive today knowing his opinion about the modern world
I'm of the Mind
music is a reflection of the human experience born from organic elements.
Believing that AI will somehow capture the complexity that lies within the human brain that is expressed through word and music is merely a fantasy.
This video is a whole ass trip
Domi JD Beck was spot on
OK my friend, please can you point me in the direction of how you created the voices used in this video, like the one in the intro. Thanks for all your hard work!
damn bruv, well done ~ I'm scared and excited to see what's next, this shit is moving so fast
i love your editing style its so good
I'll call it AI when we laugh together... Otherwise there's no soul, no emotion.
Once it can accept complicated and specific prompts with specific constraints. Then I guess ill just bow down to our robot overlords and I'll play whatever they come up with
Is Bard voiced by Scarlet Johanssen? The edits are amazing bro. Keep them coming.
Text to Chords (only for patreons): I need assistance in producing AI-generated text that I convert to music using MIDI files. Initially, I'll provide a description of the format I need for the textual representation of the music. Since music is a time-based art form, the notes follow each other in time, and sometimes there are no notes, that is, silences. Here I need you to generate Chords, so, more than one note each time, could be 2, 3, 4, etc.
The way I would like you to generate them is as follows:
Each chord is represented by these elements:
The pitches of the notes (integer values).
Because I will use this text representation and convert it to MIDI, the note should be a number from 21 (that is note A0-27.50 Hz) to 96 (that is C7-2093 Hz), so use these numbers to represent the note.
The duration of the note (float value) represented as:
0.125 for an eighth note
0.25 for a quarter note
0.5 for a half note
1 for a whole note
2 for a double whole note
But it could be any number between 0 and 2, because you know, musicians are creative, so why not 0.29 or 1.22, etc.
With this format, I need you to generate a text that I will convert into music in this format:
chords_pitch_duration_data = [ ((note, note, note), duration), ((note, note), duration), (note, duration), etc ]
And when there is a silence, the note should be 0 and the duration is how long that silence lasts.
The key is that harmony is the vertical aspect of music, or the combination of different pitches sounding at the same time. Chords are the building blocks of harmony, and they are made up of two or more notes that are played together.
The simplest type of chord is a two-note chord, also known as a diatonic chord. Diatonic chords are made up of two notes that are next to each other on the musical scale. For example, a C major chord is made up of the notes C and E.
Three-note chords are also known as triads. Triads are made up of a root note, a third, and a fifth. The root note is the lowest note in the chord, the third is the note that is two semitones above the root, and the fifth is the note that is seven semitones above the root. For example, a C major triad is made up of the notes C, E, and G.
Four-note chords are also known as seventh chords. Seventh chords are made up of a root note, a third, a fifth, and a seventh. The seventh note can be either major or minor. For example, a C major seventh chord is made up of the notes C, E, G, and B.
And you also have:
Sus chords: Sus chords are made up of a root note, a third, and a fifth, but the third is replaced with a suspended fourth or second. For example, a Csus4 chord is made up of the notes C, F, and G.
Add chords: Add chords are made up of a root note, a third, a fifth, and an additional note. For example, a Cadd9 chord is made up of the notes C, E, G, and D.
Augmented chords: Augmented chords are made up of a root note, a major third, and a perfect fifth. For example, a C augmented chord is made up of the notes C, E#, and G#.
Diminished chords: Diminished chords are made up of a root note, a minor third, and a diminished fifth. For example, a C diminished chord is made up of the notes C, Eb, and Gb.
Harmony is an important part of music, and it can be used to create a variety of different moods and emotions. For example, major chords are often used to create a sense of happiness or joy, while minor chords are often used to create a sense of sadness or melancholy.
Please note that AI-generated music may not sound pleasing as it is randomly generated. So, we will use music theory, not random math. Take into account musical concepts like scales, modes, etc.
Now that you have a full understanding of the text representation, we will create some awesome music!
Are you ready to start generating music? If so, respond with 'YES' and nothing else.
General Midi is the key of midi files
Artificial intelligence still lacks the magic that really moves the spirit. There will probably never be a way to create music the way Ryo Kawasaki does.
The once said a computer would never be able to beata grandmaster.
I love your work!! Abrazo desde Argentina!!! (detecto algun gen latino...jajaja)
Impressive as always!
As a music producer, composer and jazz pianist, i considered every one of these a certified masterpiece. We finally have access to the best music without limitations.
Pretty sure this video is being ironic.
Fantastic video as usual!!
Made it about half way thru... There's clearly enormous potential, but it's still a long way off putting Ed Sheeran outta business. I'll come back and check out your A.I. MIDI adventures in 6 months.
What software is used to replacate the voice of Scarlett ?
You are Mad, sir! Mad I tell you!!!
Whaooo! you´re so good and so fun or was it you´re så (sorry that was swedish from the pure forest mountain in Stockholm where I´m living: GPR that...) so good and so fun.
How would I know which is right? Good and fun or fun and good. Not to mention: creative to begin with... Thank you!
scarlett 🥰
awesome job... great video thanks
“timbre” is pronounced like “TAM-burr”, not like harvested trees.
what was used for the vocals?
Haha we are artist, she said. We dont code. Hahahaha. Yeah is true but ai is here to help artist with that. Great job! Nice awesome. :)
The video was great , a real art piece. The generated music was absolutely terrible , but still made for an entertaining vid with all that work put in.
Parabéns. Otimo vídeo!!!
that was in the voice of scarlet Johannsen?
Well, i persevered to the end. Music has been a significant part of my life, from rock through to deepest classical, so I'm accepting of a wide range of the world of music. Despite the hard work you put in, everything here was absolutely horrendous. None, yes, NONE, of the generated "melodies" were even remotely melodic - for any music genre, and the harmonies were similarly utterly useless. Your work on the visuals was great. Take that as a pointer to where your talents reside. But thank you for your efforts.
Of course,the vid is around 12 months old, and advances in AI music have been great in that short time, so AIVI may have improved, and it perhaps can also give better results in more musical hands. [And, text-to-music is a demanding concept in itself.]
great work!!!! thanks
Fascinating stuff
Still sounds rubbish, though. Call back when A.I. can write something as evocative and melodious as, say, Ravel's 'Piano Concerto in G Major, M. 83 - II. Adagio assai' (with full string orchestra).
I love how you used so many different ai llm tools to generate this video from a optimistic techy standpoint, but on the otherhand this video fucking hurt my brain because jesus christ the outputs were so fucking mid.
07:30 gets awesome.
Shoutout to "HER"
The reason why I will never be impressed with A.I. is because it's still just a computer functioning on 1s and 0s. Even though the data base is huge and continues to grow, it's still just a glorified calculator at the end of the day, capable of computing via language instead of numbers.
Awesome !!!
AI DESTROYS THE WORLDDDDDDDD
Like if making neo-soul chord progression is difficult.
superb!!!
Wow. Just wow.
Hi everyone, great video.
I cant make it work because I need a mido module: "ModuleNotFoundError: No module named 'mido'"
Any advice on this ?
ok i see the ;link to install mido package but I cant download it ..
Ah ok I just have to run the cell
great job man. Even noob like me can make it work
💡
Great video amd loce the experiments. But ai eill never compose better than human producers and musicias. The minor key melodies were not bad.😅
Honestly. I think the music ChatGPT created sucks
Of course. I don’t think anyone likes it. The sarcasm in this video is turned up to 11.
Yes, it does lack of life.
But it could just be his tools also, maybe just adding a longer release could solve this problem.
"Wow, this AI-Powered MIDI Generation in Advanced Mode is mind-blowing! The level of creativity and musicality it brings is simply remarkable. It's incredible to witness how AI has evolved to compose music that rivals human composers. The melodies, harmonies, and rhythms produced are truly mesmerizing. This technology has the potential to revolutionize the music industry and open up endless possibilities for musicians and artists. Kudos to the developers for pushing the boundaries of AI and music. I can't wait to see what the future holds for AI in music composition!"
Positive coment made with IA
😂😂
Not even close! Has a long way to go before it rivals a mediocre human composer!
furball
Such a beautiful video + sound + storytelling + voice over. Unbelievable. How long did it take you to make it?
Yes, I second that. Inquiring minds would like to know. 😎
This video was made by ChatGPT
The prompt: I need assistance in producing AI-generated text
that I convert to music using MIDI files. Initially,
I'll provide a description of the format I need for
the textual representation of the music.
Since music is a time-based art form,
the notes follow each other in time, and
sometimes there are no notes, that is, silences.
The way I would like you to generate them is as
follows:
Each note is represented as a tuple of two
elements:
The pitch of the note (integer value).
Because I will use this text representation and
convert to MIDI the note should be a number
from 21 (that is note A0 - 27,50 Hz) to 96 (that is
C7 - 2093 hz) so use these numbers to represent
the note.
The duration of the note (float value)
represented as:
0.125 for an eighth note
0.25 for a quarter note
0.5 for a half note
1 for a whole note
2 for a double whole note
But could be any number between 0 and 2,
bocouse you know, musician are creative so why
not 0.29 or 1.22, etc.
With this format i need you generate a text that
i will covert in music in this format:
melody_pitch_duration_data = [
(note, duration), (note, duration), (note,
duration),
etc,
]
And when there is a silence the note should be 0
and the duration is how long is that silence.
A melody is a linear sequence of notes that the
listener hears as a single entity. It is the
foreground to the backing elements and is a
combination of pitch and rhythm. Sequences of
notes that comprise melody are musically
satisfying and are often the most memorable
part of a song.
There are many ways to describe a melody. Here
are a few:
● Pitch: The pitch of a melody is the relative
highness or lowness of the notes. Melodies
can be high, low, or somewhere in between.
● Rhythm: The rhythm of a melody is the
pattern of long and short notes. Melodies can
have a slow, steady rhythm, a fast,
syncopated rhythm, or something in
between.
● Intervals: Intervals are the distance between
notes. Melodies can use a variety of
intervals, from small steps to large leaps.
● Contour: The contour of a melody is the
overall shape of the melody. Melodies can be
ascending, descending, or something in
between.
● Tonal center: The tonal center of a melody is
the note that the melody feels like it is
centered around. Melodies can have a strong
tonal center, a weak tonal center, or no tonal
center at all.
When describing a melody, it is important to
consider all of these factors. The pitch, rhythm,
intervals, contour, and tonal center all
contribute to the overall sound of the melody.
Here are some examples of how to describe
melodies:
● The melody of "Happy Birthday" is simple and
repetitive, with a clear tonal center.
● The melody of "Yesterday" by The Beatles is
more complex, with a variety of intervals and
a changing tonal center.
● The melody of "Bohemian Rhapsody" by
Queen is highly dramatic, with a wide range
of pitches and rhythms.
Quality melodies typically limit their range to
about an octave-and-a-half, feature repeating
elements like melodic intervals and rhythmic
patterns, and consist of stepwise motion with
occasional energetic leaps. Good melodies also
interact meaningfully with the bass line,
employing a mix of parallel, similar, oblique, or
contrary motions for a dynamic, counter melodic
effect. Finally, a standout melody tends to have
a climactic moment, often a high note with
significant harmonization and strong rhythmic
placement, which then descends to a restful
cadence.
No matter how it is described, a melody is one of
the most important elements of music. It is what
gives a song its identity and makes it memorable.
Please note that AI-generated music may not
sound pleasing as it is randomly generated so we
will use music theory but not random math so
don't randomize the generation process. take
into account musical concepts like scales,
modes, etc.
Now that you have a full understanding of the
text representation, we will create some
awesome music!
Are you ready to start generating music?
If so, respond with ‘YES’ and nothing else.
These Videos are so well produced and Fantastic Content .... Skynet is just a step away lol 😁, Thank You 👌
Love the way you brought all the characters from music and science to life. Has to be one of the greatest vids I’ve seen on You Tube
Wait but how did he generate the voices so well.. like in the D Angelo example. Can anyone help me with that part? Clearly that has nothing to do with the midi chords.
Blown away by how much work you put into these videos. Thanks
The video production is outstanding. Content and is great too!
YOU are a master of creativity, god bless you my friend! I´m super engaged with your videos
I am much more interested in how you got Scarlett Johansson's voice to read the text and how you got a singer to do a whole bunch of complex harmonies in the middle of the video. Where is that coming from? What apps are you using?
This is fantastic! Like an AI collage.
I would love to know which AI you're using for the talking pictures and voices.
might not be ai..possibly something like 'crazy talk' they often give out free versions of early editions. GAOTD often feature such.
That female voice is of Scarlett Johansson. She was the actress in the movie "Lucy" (2014). About a woman who became 1,000,000 times more advanced than a quantum computer. Good choice.
She was more contextually the voice of the OS in Her.
@@tonycowin Yes I had seen that a couple of times but I had almost forgotten about it when I saw this. It was good.
@@JohnNesbit1957 Yeah really enjoyed it. Johansson has done some great sci-fi. Under the Skin and The Island are my two favourites.
So, I wonder if there would be anything like the kind of prompt mashing that stable diffusion and other image generators have, where you can just throw it a whole heap of names and tell it to make a collab between a wide range of artists to get a mashup style in quite a specific way.
Artificial intelligence does not possess a mind like our own therefore cannot get inspired like Mozart did resulting in recycled outcomes.
love your vids! I can tell you used musiclm for the music at :25, it gave me an extremely similar sounding piece of music that I then used as inspiration for a piece of my own so its basically embedded into my ear. keep killing it bro
D'angelo sounded really impressive, as did Domi and JD Beck. Thank you man.
How did you get Scarlet Johansson to do the voice work?.......oh, right.
This is exactly what i was looking for. Please keep it up. Any idea how to add harmonic parts including unique styles of other instruments?
is it not easier to just use the piano roll and all the tools that DAWs already have? This seems like a ton of work.
It's more about research and technological advancement than trying to find a way out of learning/doing music production
Yes of course LOL. But once you have found a prompt that gets back better musical results, that work is done - save it and you’ll never have to make it again. If someone’s already found that prompt for you, that’s a whole lot of work you’ve skipped already. And, in the future, if somebody makes an online interface that takes your text as input, prompts ChatGPT for you, and sends the created song back to you as a MIDI file, you can now run through this whole process in a fraction of the effort and time. It’s far from being a replacement for human creativity, and it should remain that way. But at this point it is a semi-competent musical idea generator, with some actual interesting melodies/progressions among the noise it spits out. With enough time and better models, the idea generation could become more consistent and the process more streamlined. And I mean, that sounds pretty neat. I know I really enjoyed messing with MuseNet when it was online, so the idea that a general-purpose LLM could someday serve the same role is pretty wild to me.
The Beatles one was pure crap. lol
In this comment section, we have two comments from music producers. One saying trash. The other saying legit on what was spit out. So it’s not about AI, human or nature. It’s how it settles in our ears.
Carry on with whatever you are using. Someone will like/hate it. The human story.
Nine months later and all I have to do is type a sentence or two in an app called Suno and I get a fully arranged song with lyrics and vocals. Nine months from now, it will be broadcast quality, separate tracks, and we''ll have the ability to dictate the chord progressions, melody, arrangement, playing styles and singer's voice and style. Nine months after that, we're all dead because all these companies are releasing AI code as open source. Won't take long before the wrong people get it and everything starts falling apart due to crime, fraud, disinformation, political and social unrest and other nefarious uses. Open source means everyone, not just good people. BUT... In the meantime, I've loving what it's doing with music!!
Iget an error with the chords TypeError Traceback (most recent call last)
in ()
17 current_time = 0
18 for chord_pitches, duration in chords_pitch_duration_data:
---> 19 for pitch in chord_pitches:
20 midi_file.addNote(track=0, channel=channel, pitch=pitch, time=current_time, duration=duration, volume=volume)
21 current_time += duration
TypeError: 'int' object is not iterable
ChatGPT was never trained to generate music so the fact that it does anything at all is pretty remarkable. With the advent of systems specifically trained for generating music, it's clear that there's no need for sarcasm anymore, and ample room for distress among the music lovers and professionals...
Nice work, but how are you generating those singing, and talking images like D'Angelo, and The Beatles?
Great video! Plus Scarlett Johansson is a maestra.
I tried having GPT generate a number of clips in a Phrygian key. It did… okay after awhile but nothing inspiring. Eventually it gave up and told me I need to try a different algorithm.
You folks did one HELLUVALOTTAWOIK!!! Zappa's piece was pretty close to bein' sumpin'...the rest (sept4 Domi-Beck) was pretty un-inspiring. Let's hope AI doesn't get the creative love & passion that the future artists of the World put their minds to. You really saved us a LOT of time...For that, I AM TRULY grateful!!!
WOW³... not the AI output... the meticulous work you put into the video!!! How did you find and edit all these funny old clips? GREAT WORK.
ChatBots are not the right tool. AT ALL. Look, Apple already has something going on this front with its automatic musical pattern generators, even in the humble GarageBand. What’s needed is progress in developing an AI specifically designed for MIDI in a DAW, trained in a variety of music styles, and capable of being further trained by analyzing a piece of MIDI music inside a DAW. So, for example, I should be able to write two or three cues for a TV show, then have the AI tool study my three cues and then generate more cues with the same template in my style. That’s coming. It’s probably already sitting in a lab somewhere in Silicon Valley or Boston or the Rhine or Israel or who knows where.
Bard more than likely slowed down its melodies because it doesn't understand when it's looping a sequence If the next sequence doesn't have the exact same amount of ticks or if there was a duration of time where no note was played in the melody where Bard didn't count the empty space of ticks, will caused it to screw up its timing. Thus if you tell it to strictly stay with in the confines of 60 ticks per 32nd note or 1960 ticks per bar and make sure it counts any empty space as well. It should work. Or simply changing all of the 0.5 and 1.0 to 0.25 would have worked also.
It will sooner then you think. It needs a clear "understanding" of what makes us like or dislike in music, what is harmony, itc. In other words it needs to learn and it will learn much much faster. But stop, maybe we should not allow it becoming superior and just kill it while its not too late?
Well the music is far from being impressive. But I think with a more specialised corpus, better embeddings and prompt engineering considering other dimensions (dynamics, timbre, range, music forms, motive, transition/ponctuation...), it will lead to more exploitable material.
It's just the beginning anyway. It's already transforming sound engineering, it will sound transform actual music production. I hope humanity will still enjoy create and perform music anyway...
.
Toca como alguien muy experimentado, el problema es que no a toda la gente de la generación actual le gusta la música complex, la mayoría de estos casos es música para músicos, porque si le pones una 7ma de ebmaj, y empiezas de allí, hay mucha gente que nos sorprenden los acordes aumentados y otros que no, a lo que voy es que si te ayuda la ia no te dará el super trabajo 2023 que revolucionaras la industria, solo sacaras música como si fueses un músico experimentado que siento que es el peor engaño que te puedes hacer a ti mismo. Si algun día tocas en un concierto no creo que pongas el pc con la ia para que te toque el pianito, una cosa es saber utilizarlo bien y otro que te haga todo el trabajo que en el fondo te quita merito propio
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
Turn off your computers, take your instruments and make noise. Don't play video games, play BbMaj7. Good God, Real Music By Real Musicians never die. 💜☔️
you have so many cliche, over-used words and phrases that after 2 minutes of listening to you, I couldn't take it any more and disconnected. Consider backing off these worn out spoken words and using simple instead. More message, less ornamental BS
as a music producer, composer, and jazz pianist nearly all of melodies were like literally terrible. sure, they sound decent in some places but there's no structure and none of that human "feel" that actually makes music be a reasonable display of emotion. looks like a fun tool to mess with but in no way ready for production ready music
Almost as if it’s beta
@@finlay1702it’ll never be able to come close to a talented human. Personally, when I listen to music I always think wow there’s a human behind this what a creative dream they’ve let me into but obviously if you use ai it won’t have any of that
😂
@@danielkillorin9742 AI will never create Art. I've heard that before 😅
Many actual musically talented humans have difficulty using music as an expressive medium, how well do you think something without feelings will be able to do it?
The massive problem with chat gpt or any other ai is the welfare sized token alottment that you get. Even with gpt4 its only 18,000 tokens . Once your code gets to a certain size the AI simply gets chronic amnesia and your progress comes to a sceeching halt as you run around in circles trying to solve problems while the ai continuosly loses its memory with everytime you post the new scrypt. Claude AI boasts a nice 100,00 tokens but its coding skills and unwillingness to complete anything makes it pretty much useless.
Honestly... All those melodies suck... and hopefully it stays that way... The last thing we need is some fake soulless music created by algorithms..