As a music producer, this tool is 100% a W for people to help themselves get over a block. At the same time I hope that traditional methods still remain more utilized so we retain character through different artists. I could see myself using this to get an idea of something to do in a track, then making something inspired by the AI-generated stems. Dope program
Yeah, yeah. As a music producer this is a 100% win for anyone that doesn't want to pay artists. It's just a new form of piracy - train the model on the humans and then forget about them.
i agree, i had a creative block, and just used suno for some quick sampling ,,, BRO it was so fun I got the divine blessings to move forward with the project with fresh ideas
I looked for this and during my search found that many others were interested in the same thing - an AI tool that will RErecord a song with new lyrics that the user provides. The interest in this is huge. You're welcome.
I do like this but I believe the future is Audio Upscaling. You just rough something you on the guitar, you beat box some drums, use a pitch changer to turn the guitar into a bass, and record it all on your iPhone 5 microphone. Then you upload the tracks and the AI makes it all studio quality. Writing songs is the awesome part. The sucky part is doing it over and over trying to get it perfect, and then doing all the mixing to make it sound good. This sort of thing isfor people who can't write songs, and prefer to just do the final mixing part. Which is all well and good but I feel like most people have ideas they want made into awesome things. Not wanting to turn AI ideas into awesome things.
This is something these types of models (Diffusion) can do for free (to some degree). It kinda works like a subtractive synthesizer: you start from a pure noise signal, and you iteratively refine it until it looks like something out of your training data. Instead of starting from pure noise, you can start from your "rough something," add to it a bit of noise to mask out details, and ask the generative model to regenerate the missing stuff. This way one can start from hummed melody, and transform it into a piano. Or from a beatbox to drums, etc etc We haven't tested it extensively in the case of Diff-A-Riff, but I'm sure it can already do some fair job at this
I pretty much agree, but if you're a real musician-which I think might be dying away-you don't have to do it over and over again, and you're not after perfection anyway. Real musicians do not achieve perfection.
Sounds like you are a lazy musician that doesn’t like to practice and look down on people making a song sound great. As if it is a 2nd grade job to do when you can’t write. You don’t know anything about production and the skills it takes. You will suck at production if you don’t know how to write. AI is a blessing and horror at the same time. Urghhh, you prove what’s coming at us…a whole lot of crap by lazy people
If you're an advanced Reaper user, you can already do it since several months using 'Reaper Composer Assistant'. The model is trained only on royalty-free data ( so of course it's not very "modern", but it still allows you to find inspiration very quickly ). There are a ton of midi plugins without AI that allow you to do more or less the same thing though (Reacomposer is one that is free under Reaper, RapidComposer is paid and little known but probably the most complete, etc). I have tested a lot of things, but for now I think that HI (Human Intelligence) seems even more inspired than AI ;=)
Anime related to AI DJs and AI Music: 1. Carol and Tuesday 2. Macross Plus Both excellent and directed Shinchiro Watanabe (Cowboy Bebop and Samurai Champloo)
Nice video! Subscribed! It sounds a bit like the late 90s downloads with 32kbps mp3s. But the creations of the AI is impressive. In the 90s, many people thought that AI couldn't be creative.
"this sucks, my family will starve now." This is what artists said about photography, what musicians said about synths, what singers said about autotune, and what people are currently saying about ai, they all have one thing in common, the new hasn't killed the old.
As a real music producer I can assure that this sounds bad. The separate tracks or instruments do not harmonize together at all. It is like hearing a singer singing horribly false. This would certainly never make a hit song. There are some other AI apps though that can produce pretty crazy additions to songs that still need human and professional producers to make hit songs that people can enjoy. The AI stuff mostly still have an eerie sound to it. Peace out! ❤
I think for demos and writing this will be excellent for singers/rappers who can’t afford pre production costs. I’m happy it isn’t pumping out banger that are totally mixed and mastered yet. I’ve heard so many horror stories of women working with producers who are perverts so this would bypass all that which is a w. But yes I agree it’s not quite understanding the harmonies and I’m hearing lots of strange artifacts.
It's just a demo, not how to make the next hit song. This AI aims to simplify music production and expand creative possibilities, not to replace the role of a music producer. Melancholic pads and blues bass might not blend well and could be even in different keys. Not a producer
This can be used by real musicians by just getting the song completed using whatever AI you need and then just re-recording the tracks by real musicians. Mute or erase the AI generated so-called music..
Super cool. 👌Thanks for sharing! 🤗 I just wish that someone could develop something similar, but generating midi tracks instead of low-fi audio stems. I could of course use melodyne (build into Cakewalk) to transcribe the generated audio to midi data. But still I would really love to have it generate midi data directly.
You’ve been able to do that for decades with samples and arpeggio creators in DAWs. Also, there are MIDI kits with pre written melodies and chord progressions that can be modified. Nothing new here. The technology has just gotten more advanced
@@Author_SoftwareDesigner Those methods required substantially more effort, proficiency and musicality than just having to repeatedly click retry to obtain a unique and serviceable result. So there's definitely something new here.
@@c0ldc0neyeah they love to brag they don’t need any music knowledge and click the mouse and the computer does it all. Probably why so much of this stuff sounds the same bland generic stuff I’m self taught but seem to write better now I’ve learnt a bit of music theory and why certain chords work
People saying this sounds bad.. No. This is great for bedroom producers with limited resources and little to no money. It makes a perfect foundation to build from. Perfect for demonstrations. It's better than being stuck uninspired because you don't have such instruments, or can't play them well enough to get inspired from. They can even use this audio to feed into Suno or Udio, to build from even more. Once producers have the skill-set and resources they can re-produce tracks they made with the help of Ai, in higher-quality, and humanise it. Every big track starts out with a demo, many of them sound pretty rough but the vision is there being communicated, and that's all that's needed in order to turn it into a top teir sounding track later down the line. If Ai can help people lay out their idea's well enough to inspire great sounding records later down the track, fantastic. If nothing comes from it, so? Atleast you're trying and spending time doing something you like, likely learning new things. Ai isn't about replacing producers and song-writers it's about giving them more resources and access to samples they need to execute their creative vision.
I see your point about using AI for inspiration and see nothing wrong with that, and if someone can afford something like the Korg NTS-1 further down the line then loads it up with free custom oscillators and effects, they can make pretty much any sound they can possibly imagine including drums and percussion in very close to studio quality. I've been doing this for a couple of months now sampling, composing and mixing in Ardour (a free multi-platform DAW) on a Raspberry Pi5 with a budget ADC/DAC HAT, and the end results sound massively better than this software seems able to achieve at present. A setup like mine would set someone back around £200, but that's an incredibly cheap price to pay for the almost infinite possibilities it opens up once the creative juices start flowing if electronic music is their thing.
@ I think the main issue is people get stuck with creativity. They may have the sounds but not the skill set to know how to best use them. Ai can help them with over coming that hill. Use Ai tracks as foundation and then use those machines over the top. To me, the most valuable part of Ai is finally having access to the human voice as an instrument. Producing beats and instrumentations has never been attractive to me, WITHOUT a human voice. I wanna write songs with lyrics, not lay beats. If I make a beat I want it to dance with a voice. Producers have mostly depended on synthesisers to be their voice because they themselves couldn’t sing. Ai changes that and that should excite producers. We just unlocked the ultimate instrument. The ultimate sample generator. Thus, Ai should also result in the prices of plugins and such instruments to lower. It’s actually ridiculous how expensive plugins and VSTs can get.
@bcrox that’s very reductionist and snobby. There is no one way to use Ai, how much work and creativity one puts into a project varies across projects and people. Ai is a very dynamic tool that allows you to exert either full creative control, or minimal. Plus, how many artists on the radio do you think are laying down all the stems on their tracks? How many producers even do that? Most of the best ones get musicians in and record them for samples. No one cares except gatekeepers in denial that people doing less are managing to make better music than them. Lol. Work smarter not harder.
The fact that you choose a lo-fi style combines with the style of sound the ia generated... But what if you choose a tight groovy style? The results would be good enough?
as a music producer: this is the end....because this is the pawlow sugar button that lets everyone make perfect music...so in a way its the democratization of talent, but at the same time it is also the invention of arbitrariness with absolute quality. Its like everyone can mix their own "lucky pill" now. Producers are not needed soon at all...Live musicians otherwise will still be needed..and stars
@@NostalgicMem0ries For the mediocre people, it's scary, but for those who know they can take advantage of it as a tool to leverage what they already know.
@@DrayNoR1 now yes, i agree, but in time those jobs might be obsolete if we have advanced enough AI programs with few prompts giving you what you want. Look how many coding jobs are done cause of AI in last year... many also said it will only help programmers and coders, now one dude does job of 10 20 ...
Powerful tool that will further develop ...current popular music and songwriting for demos could make use of it however putting orchestrations together may prove to be more challenging
Its the end. bc, if something is too easy, is: a. cheap b.not fun anymore c.not worth the effort. This is psychology 101, we as humans don't do things because they are easy.. cheers..
@@GustavKnuttsenshow me the AI that can make Not dog shit music show me the statistics of music producers / composers in 1924 and in 2024 relative to the global population. then we will see if your statements are true or just bullshits (1: "you can make good music with AI and composers are useless" 2: if something gets easier people dont do it anymore [music production got exponentially Easier and cheaper over the years and a godzillion more people are releasing music now])
I've had conversations with AI asking that when it takes over to wipe out the billionaires first and the rest of us will be just fine. I'm not joking. I've had this conversation.
As a music producer for hobby I'D LOVE to have something like that, especially integrated in the DAW to quickly convert tons of drafts and ideas into complete songs, even if they are low quality (I'll never get famous anyway so..).
@@276- "even if they are low quality" I mean if you wanna bury your shit in the never ending stream of slop by all means... More for others that actually care.
What's the quality of the output audio samples? It sounds quite low quality and fuzzy / noisy - which is the problem with most audio AI tools out there 😐
I actually love this idea. I always hear melodies in songs that I wish I came up with and now i'd be able to easily generate variants based off of that that are similar but just different enough that I can turn it into my own thing.
So, all we are missing now is the voice that sings the lyrics. Do they plan on adding that? Also I doubt the stitching of the different parts. How to deal with different sections in a song? Can it also generate the base layer based on a text input for the style?
SO happy that I don't need a crutch like this to make music. An excellent tool for creative people who lacks creativity and want someone else to do the work.
No way man I was just thinking yesterday how cool would it be to go back to the jam sessions with the boys back in the day. Can’t wait to get AI that can jam with you anytime any style. Huge W 🔥🔥🔥
As a documentary producer with some but not expert musical skills, this is the way forward for me to generate simple backing tracks that are more than good enough to illustrate the subject of the film.
What's wild about Suno is it's already handling the mixing and layering. Soon, you'll be able to say, 'Give me a psychedelic drum with these effects,' and just build your track from scratch. That’s gonna be so rad! 🔥🎶
This can be used by real musicians by just getting the song completed using whatever AI you need and then just re-recording the tracks by real musicians. Mute or erase the AI generated so-called music..
I like this idea because it seems to give you much more freedom and control than Suno and Udio. But also more time consuming. I suppose they all have their use cases. I'd probably end up using two of them together. Thanks for sharing! Can't wait to try it myself.
Nice to see Cakewalk by Bandlab being used. No one ever seems to talk about it. I certainly love Cakewalk by Bandlab. And - thank you for this educational video.
would love to get my hands on this to help hint at possible alt tunes to what I've created using certain instruments on parts of tracks I've created, but I wouldn't use it as an online service, no way I would trust copyrighted content uploaded and used as and how they see fit, which would undoubtedly require us to agree with some small print in some hidden t&c's somewhere... Locally run, no phone home or upload and needs to use the gpu, not cpu.
It's weird that at 10:50 it generated a synth with some bleed of the base track... That's the future of AI in DAWs, anyway, say goodbyr to the 50 terabyte sample libraries you have :)
There's a great deal of potential, which I see mainly in the arrangements, bass, guitar, drums (but with breaks and variations), etc. But only on the principle of operation, the sound still needs improvement. It sounds like old magnetic tape with chorus on it, and the sounds don't blend well. We need to leave the sounds to the musician and generate Midi. We also need to explore the idea of arranging according to a particular style (for example, horns in a salsa style, or synth layers in a pink floyd style...). In short, the principle is brilliant and it's the future, because we don't all have armies of arrangers at our service, and it will speed up production time while improving track quality. But there's still a long way to go to reach this ideal. But let's not be too harsh on these early versions: they're the beginnings of AI, and in ten years' time, those who laughed at them won't be laughing at all...
What music software is that? Looks very simple to use. FL Studio gets tiring after a while with all the clicking around. Need AI to just optimize things to a few clicks by hearing and suggesting.
@@theAIsearch I might need that. It's more catered to the musician rather than the technician sort of deal. Direct to the point, and optimal options which means more work done than the slap new feature and find ordeal in FL Studio it seems. I'll learn more into it when I can, but so far it's looking like a gem I would've liked existed before getting into FL Studio.
It sounds great however you can already do something similar in WA productions “instacomposer 2” with midi using you own sounds. What I like about this one you’re showing you can make a new audio pattern based on the sound you feed it so will look into this if it’s available
The bass and guitar sounds kind of wobbly. I don’t know if that’s because you were going for sort of Lofi feel or if the quality of the app needs improvement. Drums sound good though.
watched the end well that sounds nice. so one should make parts like this then using the circle of 5ths make other parts then put them together to make a better song.
I'm just a passerby, but diff-a-riff cannot distinguish the differences of harmonic content, much like other processes that work with the 1s and 0s of a binary output a digital signal uses to process audio. the harmonic content of a guitar or other analog gear that takes into the detail of a human quality, is the even harmonics. the odd harmonics and even harmonics are not balanced in most of these results, because the 1s and 0s cannot run a process like Fast Fourier Transform or FFT. This can work for people who want to make demos, or give rough ideas to form, but the harmonic content not balanced, at least from my own understanding of the human condition, will not pass for pleasable music to the major audience, which should include, commercial work.
Fantastic idea. I'll keep a keen eye on its release. Also, you created a beautiful tune. I couldn't help myself, so I had to import it to Suno with my own AI-modeled vocals and it turned into a beautiful song (with the 80s radio quality of course 😄) Sadly, I am the only one who can enjoy it.
It's only a matter of time before more and more and more and more AI generated music, including every instrument and every vocal, will start to dominate what people listen to.Hopefully I'm wrong about dominate.
The concept is cool. A true musician wouldn’t need this though. But for home recordings, it can be helpful for lazy producers. Now if it was real-time, that would be cool, so you can improvise with a band at home in your bedroom. As a musician, for writing, I like the part of coming up with and writing all the instrument parts, but for improv jamming, it would be awesome (if it could be real-time). For amateurs, producers who don’t play instruments or musicians who only know one instrument and probably use loops to fill space, it can be helpful. Speaking of loops, loops pretty much do the same thing but sound way more mastered. I like the comments about it doing this in midi. So the sound quality wouldn’t be an issue. So, doing it in midi, in real-time, following you while recording, now you’ve got a winner.
This is crazy. I'm curious if it could be combined with a tool like MicroMusic, since it turns any sound into a synthesizer preset. If you'd like to try it I can give you a free license / affiliate link
Excellent video, I love that it doesn't replace the artist like udio, but is a tool. The title is not clickbait, this workflow truly looks like the future of music production.
Like Output coproducer but much more to the point, at least for my uses. Useful when I have a certain sound in mind but not a sample or instrument that is quite what I want. The guitar strumming with a big better quality could be excellent.
As a music producer, this tool is 100% a W for people to help themselves get over a block. At the same time I hope that traditional methods still remain more utilized so we retain character through different artists. I could see myself using this to get an idea of something to do in a track, then making something inspired by the AI-generated stems. Dope program
Agree!
Just use Spice to get over block, making music isn't difficult
Yeah, yeah. As a music producer this is a 100% win for anyone that doesn't want to pay artists. It's just a new form of piracy - train the model on the humans and then forget about them.
i agree, i had a creative block, and just used suno for some quick sampling ,,, BRO it was so fun I got the divine blessings to move forward with the project with fresh ideas
@@nagoranerides3150 don't cry, this is the future
The Guitar sound is the one that convinced me, for me is really difficult to do a convincing strum on a vst, can't wait to use it
Strum on a guitar maybe?
@@tomm5023some people don't own guitars.
I looked for this and during my search found that many others were interested in the same thing - an AI tool that will RErecord a song with new lyrics that the user provides. The interest in this is huge. You're welcome.
I do like this but I believe the future is Audio Upscaling. You just rough something you on the guitar, you beat box some drums, use a pitch changer to turn the guitar into a bass, and record it all on your iPhone 5 microphone. Then you upload the tracks and the AI makes it all studio quality. Writing songs is the awesome part. The sucky part is doing it over and over trying to get it perfect, and then doing all the mixing to make it sound good. This sort of thing isfor people who can't write songs, and prefer to just do the final mixing part. Which is all well and good but I feel like most people have ideas they want made into awesome things. Not wanting to turn AI ideas into awesome things.
that's a great idea!
This is something these types of models (Diffusion) can do for free (to some degree). It kinda works like a subtractive synthesizer: you start from a pure noise signal, and you iteratively refine it until it looks like something out of your training data. Instead of starting from pure noise, you can start from your "rough something," add to it a bit of noise to mask out details, and ask the generative model to regenerate the missing stuff. This way one can start from hummed melody, and transform it into a piano. Or from a beatbox to drums, etc etc
We haven't tested it extensively in the case of Diff-A-Riff, but I'm sure it can already do some fair job at this
I pretty much agree, but if you're a real musician-which I think might be dying away-you don't have to do it over and over again, and you're not after perfection anyway. Real musicians do not achieve perfection.
Sounds like you are a lazy musician that doesn’t like to practice and look down on people making a song sound great. As if it is a 2nd grade job to do when you can’t write. You don’t know anything about production and the skills it takes. You will suck at production if you don’t know how to write. AI is a blessing and horror at the same time. Urghhh, you prove what’s coming at us…a whole lot of crap by lazy people
@@bluzedogg AMEN BRO!
I think being able to have a midi option would make it better. That way you can edit the midi notes as you please.
thatd be nice
SADLY this does not work
cant you use melodyne
@@klutzershyy any type of pitch correction will make it sound objectively worse.
I prefer AI that creates MIDI from MIDI files and lets me choose any sound from my sound library
If you're an advanced Reaper user, you can already do it since several months using 'Reaper Composer Assistant'. The model is trained only on royalty-free data ( so of course it's not very "modern", but it still allows you to find inspiration very quickly ). There are a ton of midi plugins without AI that allow you to do more or less the same thing though (Reacomposer is one that is free under Reaper, RapidComposer is paid and little known but probably the most complete, etc).
I have tested a lot of things, but for now I think that HI (Human Intelligence) seems even more inspired than AI ;=)
AIVA, Ideogram
What you've just described is a sound engine. You should try something like FL Studio or Ableton:D
@@kotovasiya-tv That is not at all what they just described
@AlenNoble I don't mean a DAW I mean an AI generator that creates midi from different MIDI files to create a new different one
Anime related to AI DJs and AI Music:
1. Carol and Tuesday
2. Macross Plus
Both excellent and directed Shinchiro Watanabe (Cowboy Bebop and Samurai Champloo)
Nice video! Subscribed! It sounds a bit like the late 90s downloads with 32kbps mp3s. But the creations of the AI is impressive. In the 90s, many people thought that AI couldn't be creative.
"this sucks, my family will starve now." This is what artists said about photography, what musicians said about synths, what singers said about autotune, and what people are currently saying about ai, they all have one thing in common, the new hasn't killed the old.
As a real music producer I can assure that this sounds bad. The separate tracks or instruments do not harmonize together at all. It is like hearing a singer singing horribly false. This would certainly never make a hit song. There are some other AI apps though that can produce pretty crazy additions to songs that still need human and professional producers to make hit songs that people can enjoy. The AI stuff mostly still have an eerie sound to it. Peace out! ❤
For sure true. For people who aren't able to sing or play an instrument, it's a great help get going in this space. 😃
thats good ...now i feel really assured
I think for demos and writing this will be excellent for singers/rappers who can’t afford pre production costs. I’m happy it isn’t pumping out banger that are totally mixed and mastered yet. I’ve heard so many horror stories of women working with producers who are perverts so this would bypass all that which is a w. But yes I agree it’s not quite understanding the harmonies and I’m hearing lots of strange artifacts.
It's just a demo, not how to make the next hit song. This AI aims to simplify music production and expand creative possibilities, not to replace the role of a music producer. Melancholic pads and blues bass might not blend well and could be even in different keys. Not a producer
@@BikeAreathen they should dedicate their time to something useful 😅
the tech is very impressive, and I know it will improve, but the end result did not sound good to me at all 🤷
This can be used by real musicians by just getting the song completed using whatever AI you need and then just re-recording the tracks by real musicians. Mute or erase the AI generated so-called music..
WOW man, that's actually pretty awesome... I want it now!!!!! Thanks for Sharing!
love this experimental stuff!
Super cool. 👌Thanks for sharing! 🤗
I just wish that someone could develop something similar, but generating midi tracks instead of low-fi audio stems.
I could of course use melodyne (build into Cakewalk) to transcribe the generated audio to midi data. But still I would really love to have it generate midi data directly.
Now you can call yourself a music producer without being able to produce any music.
Brave new world
The mediocre will inherit the earth
You’ve been able to do that for decades with samples and arpeggio creators in DAWs. Also, there are MIDI kits with pre written melodies and chord progressions that can be modified. Nothing new here. The technology has just gotten more advanced
@@Author_SoftwareDesigner Those methods required substantially more effort, proficiency and musicality than just having to repeatedly click retry to obtain a unique and serviceable result. So there's definitely something new here.
@@c0ldc0neyeah they love to brag they don’t need any music knowledge and click the mouse and the computer does it all.
Probably why so much of this stuff sounds the same bland generic stuff
I’m self taught but seem to write better now I’ve learnt a bit of music theory and why certain chords work
People saying this sounds bad.. No. This is great for bedroom producers with limited resources and little to no money. It makes a perfect foundation to build from. Perfect for demonstrations. It's better than being stuck uninspired because you don't have such instruments, or can't play them well enough to get inspired from. They can even use this audio to feed into Suno or Udio, to build from even more. Once producers have the skill-set and resources they can re-produce tracks they made with the help of Ai, in higher-quality, and humanise it. Every big track starts out with a demo, many of them sound pretty rough but the vision is there being communicated, and that's all that's needed in order to turn it into a top teir sounding track later down the line. If Ai can help people lay out their idea's well enough to inspire great sounding records later down the track, fantastic. If nothing comes from it, so? Atleast you're trying and spending time doing something you like, likely learning new things. Ai isn't about replacing producers and song-writers it's about giving them more resources and access to samples they need to execute their creative vision.
I see your point about using AI for inspiration and see nothing wrong with that, and if someone can afford something like the Korg NTS-1 further down the line then loads it up with free custom oscillators and effects, they can make pretty much any sound they can possibly imagine including drums and percussion in very close to studio quality.
I've been doing this for a couple of months now sampling, composing and mixing in Ardour (a free multi-platform DAW) on a Raspberry Pi5 with a budget ADC/DAC HAT, and the end results sound massively better than this software seems able to achieve at present.
A setup like mine would set someone back around £200, but that's an incredibly cheap price to pay for the almost infinite possibilities it opens up once the creative juices start flowing if electronic music is their thing.
@ I think the main issue is people get stuck with creativity. They may have the sounds but not the skill set to know how to best use them. Ai can help them with over coming that hill. Use Ai tracks as foundation and then use those machines over the top. To me, the most valuable part of Ai is finally having access to the human voice as an instrument. Producing beats and instrumentations has never been attractive to me, WITHOUT a human voice. I wanna write songs with lyrics, not lay beats. If I make a beat I want it to dance with a voice. Producers have mostly depended on synthesisers to be their voice because they themselves couldn’t sing. Ai changes that and that should excite producers. We just unlocked the ultimate instrument. The ultimate sample generator. Thus, Ai should also result in the prices of plugins and such instruments to lower. It’s actually ridiculous how expensive plugins and VSTs can get.
@bcrox that’s very reductionist and snobby. There is no one way to use Ai, how much work and creativity one puts into a project varies across projects and people. Ai is a very dynamic tool that allows you to exert either full creative control, or minimal. Plus, how many artists on the radio do you think are laying down all the stems on their tracks? How many producers even do that? Most of the best ones get musicians in and record them for samples. No one cares except gatekeepers in denial that people doing less are managing to make better music than them. Lol. Work smarter not harder.
The fact that you choose a lo-fi style combines with the style of sound the ia generated... But what if you choose a tight groovy style? The results would be good enough?
As a Music Producer , this is amazing
😃
more scary than amazing, in few tears this will be so advanced that it will require only prompt and another job will be gone :D
as a music producer: this is the end....because this is the pawlow sugar button that lets everyone make perfect music...so in a way its the democratization of talent, but at the same time it is also the invention of arbitrariness with absolute quality. Its like everyone can mix their own "lucky pill" now. Producers are not needed soon at all...Live musicians otherwise will still be needed..and stars
@@NostalgicMem0ries For the mediocre people, it's scary, but for those who know they can take advantage of it as a tool to leverage what they already know.
@@DrayNoR1 now yes, i agree, but in time those jobs might be obsolete if we have advanced enough AI programs with few prompts giving you what you want. Look how many coding jobs are done cause of AI in last year... many also said it will only help programmers and coders, now one dude does job of 10 20 ...
Does "drunk chipmunks humming a Christmas melody" work for a text-to-instrument..?
Powerful tool that will further develop ...current popular music and songwriting for demos could make use of it however putting orchestrations together may prove to be more challenging
Its the end. bc, if something is too easy, is: a. cheap b.not fun anymore c.not worth the effort. This is psychology 101, we as humans don't do things because they are easy.. cheers..
we humans do things because it is easy, that is evolution 101.
@@potffin have you ever tried to create any art?
yes thats why since cameras were invented nobody draws and paionts anymore, so doom
@@bWWd0 show me the machine that can paint watercolours
@@GustavKnuttsenshow me the AI that can make Not dog shit music
show me the statistics of music producers / composers in 1924 and in 2024 relative to the global population.
then we will see if your statements are true or just bullshits
(1: "you can make good music with AI and composers are useless"
2: if something gets easier people dont do it anymore [music production got exponentially Easier and cheaper over the years and a godzillion more people are releasing music now])
How to get this?
Is there an AI that separates all the tracks of a music already created ?
just search for "stem splitter". There are a bunch of them out there.
there are a lot - like ripdaw or ultimate vocal remover
@@Vastralix womp womp
@@NCLDMR "THAT'S THE SOUND OF DA BULLIES, THAT'S THE SOUND OF DA BEAST"
Ultimate Vocal Remover.
It's free and it works locally.
Find some video for the installation because you are required to download the model separetly.
Please Ai do my taxes and let me do the creative stuff
💯
I've had conversations with AI asking that when it takes over to wipe out the billionaires first and the rest of us will be just fine. I'm not joking. I've had this conversation.
As a music producer for hobby I'D LOVE to have something like that, especially integrated in the DAW to quickly convert tons of drafts and ideas into complete songs, even if they are low quality (I'll never get famous anyway so..).
so youre basically just participating on trying to swamp the internet in low quality trash? put in some effort man.
@@HiUnidentified get a grip on it, did you not hear anything they said? 😭
@@276- "even if they are low quality"
I mean if you wanna bury your shit in the never ending stream of slop by all means... More for others that actually care.
@@HiUnidentified cry more, cry harder.
@@HiUnidentified Lol looks like somebodys pissed their job got taken 😂Cry
What's the quality of the output audio samples? It sounds quite low quality and fuzzy / noisy - which is the problem with most audio AI tools out there 😐
Oh this might be really great for making background music for my TH-cam videos.
Is it not available as an API or something yet?
not available yet. they are still developing it
I actually love this idea. I always hear melodies in songs that I wish I came up with and now i'd be able to easily generate variants based off of that that are similar but just different enough that I can turn it into my own thing.
So, all we are missing now is the voice that sings the lyrics. Do they plan on adding that? Also I doubt the stitching of the different parts. How to deal with different sections in a song? Can it also generate the base layer based on a text input for the style?
SO NICE!!! Thanks for sharing.
Thanks for watching!
What did you do to make the rhodes piano chords? They are beautiful and have a tape feel. As a non guitar player, the guitar part is sick!
SO happy that I don't need a crutch like this to make music.
An excellent tool for creative people who lacks creativity and want someone else to do the work.
No way man I was just thinking yesterday how cool would it be to go back to the jam sessions with the boys back in the day. Can’t wait to get AI that can jam with you anytime any style. Huge W 🔥🔥🔥
thanks for sharing!
As a documentary producer with some but not expert musical skills, this is the way forward for me to generate simple backing tracks that are more than good enough to illustrate the subject of the film.
Very good explanatory run-through. Thanks.
You're welcome!
How will I use it? We want to use it as you used it
it's not out yet. still in dev. when i get more info, ill announce it on my channel
@@theAIsearch ok Bro . thank you very very Much
What's wild about Suno is it's already handling the mixing and layering. Soon, you'll be able to say, 'Give me a psychedelic drum with these effects,' and just build your track from scratch. That’s gonna be so rad! 🔥🎶
2:25 Can You (or someone else) Provide a link to the page visible on the screen? ta 2:25 I just can't find it by myself...
As a multi instrumentalist and music producer, I’m terrified and also trying it out immediately.
Awesome, I've been looking for a tool like this for a while now.
This type of A.I is more of what I appreciate
It seems like we never get to choose change, only how we change as a result of it.
link?
this is craaaaaaaaaaaaazyyyyyyy a powerfull tool
Please make more, I want to see what else it can do. They had some good demos
I'd like to try a more complex chord harmonization, perhaps taking a standard tune from the real book.
That's pretty rad. I'm looking forward to trying it out.
As a “music producer” this sounds sh#t..why is it timestretching the uploaded sound & not generate a new clean instrument with a similar tone.
same thoughts. New Ableton has more advanced AI features and opportunities
This can be used by real musicians by just getting the song completed using whatever AI you need and then just re-recording the tracks by real musicians. Mute or erase the AI generated so-called music..
You don't need to be a producer to realize this, but this feature is obviously in early stages. AI music has come a long way in the last year
I noticed that in Controls section quality was set to 30. Maybe it would sound better with 100. Have you also played with that section?
I like this idea because it seems to give you much more freedom and control than Suno and Udio. But also more time consuming. I suppose they all have their use cases. I'd probably end up using two of them together. Thanks for sharing! Can't wait to try it myself.
let's hope this is out soon. can't wait to try this
Great video, I'll be interested to see it when it is ready for release along with the cost.
Non quantised drum sound might be a use case.... I hope.
Great to see Sony doing such thoughtful products.
Nice to see Cakewalk by Bandlab being used. No one ever seems to talk about it. I certainly love Cakewalk by Bandlab. And - thank you for this educational video.
You're welcome! Yeah Cakewalk is awesome
It's perhaps future for some type of music production.
amazing!looking forwards to try it!
Any idea when this might be released?
would love to get my hands on this to help hint at possible alt tunes to what I've created using certain instruments on parts of tracks I've created, but I wouldn't use it as an online service, no way I would trust copyrighted content uploaded and used as and how they see fit, which would undoubtedly require us to agree with some small print in some hidden t&c's somewhere... Locally run, no phone home or upload and needs to use the gpu, not cpu.
It's weird that at 10:50 it generated a synth with some bleed of the base track...
That's the future of AI in DAWs, anyway, say goodbyr to the 50 terabyte sample libraries you have :)
yes omg those kontakt libraries take up so much space
Sounds like I'm stuck in a lo-fi underwater elevator, very underwhelming.
I wonder if you could use it on vocals.
Where can I try this tool?
Your showcase videos are excellent! And this new AI is amazing, so much potential. Love this.
Thanks!
Hello. How can I change the Track to have copyright in youtube but to be very close to the original???
Any idea when Diff-A-Riff becomes available to the public?
People will forget how to do anything.
That is the purpose 😎
That’s probably the idea, make everyone reliant on tech
There's a great deal of potential, which I see mainly in the arrangements, bass, guitar, drums (but with breaks and variations), etc. But only on the principle of operation, the sound still needs improvement. It sounds like old magnetic tape with chorus on it, and the sounds don't blend well. We need to leave the sounds to the musician and generate Midi. We also need to explore the idea of arranging according to a particular style (for example, horns in a salsa style, or synth layers in a pink floyd style...). In short, the principle is brilliant and it's the future, because we don't all have armies of arrangers at our service, and it will speed up production time while improving track quality. But there's still a long way to go to reach this ideal. But let's not be too harsh on these early versions: they're the beginnings of AI, and in ten years' time, those who laughed at them won't be laughing at all...
how can I use it ? Where is this Interface to find? A Plugin ? an App for iPad ??
This will be MASSIV in the produser world. Make a DAW plugin of this!!!
What music software is that? Looks very simple to use. FL Studio gets tiring after a while with all the clicking around. Need AI to just optimize things to a few clicks by hearing and suggesting.
i used cakewalk. quite similar to fl studio
@@theAIsearch I might need that. It's more catered to the musician rather than the technician sort of deal. Direct to the point, and optimal options which means more work done than the slap new feature and find ordeal in FL Studio it seems. I'll learn more into it when I can, but so far it's looking like a gem I would've liked existed before getting into FL Studio.
I NEEDED A DEMO PLEASE
It sounds great however you can already do something similar in WA productions “instacomposer 2” with midi using you own sounds. What I like about this one you’re showing you can make a new audio pattern based on the sound you feed it so will look into this if it’s available
is there an ai that separates stacked harmonies into multiple stems?
Wow! 😮
Any idea when it will be available to the public?
best question 🧐
I wonder too
its still in dev and there's no timeline yet. when i get more info, ill announce it on my channel
Link to the actual tool?
Where do I get this music site?
Yo man, jow can I get access to Diff a Riff? Thnx
Hey man, your uvr5 video get deleted, any change you can upload it again?
The bass and guitar sounds kind of wobbly. I don’t know if that’s because you were going for sort of Lofi feel or if the quality of the app needs improvement. Drums sound good though.
Nice to see someone using Cakewalk!
Doesn't Audacity have something like this now using referencing to make stems? I thought I read that somewhere
this is more what we need!!!
where is the link to hugging face?
watched the end well that sounds nice. so one should make parts like this then using the circle of 5ths make other parts then put them together to make a better song.
I'm just a passerby, but diff-a-riff cannot distinguish the differences of harmonic content, much like other processes that work with the 1s and 0s of a binary output a digital signal uses to process audio. the harmonic content of a guitar or other analog gear that takes into the detail of a human quality, is the even harmonics. the odd harmonics and even harmonics are not balanced in most of these results, because the 1s and 0s cannot run a process like Fast Fourier Transform or FFT. This can work for people who want to make demos, or give rough ideas to form, but the harmonic content not balanced, at least from my own understanding of the human condition, will not pass for pleasable music to the major audience, which should include, commercial work.
when this coming out i want to use this so bad i got unfinished songs to do i see endless possibilities using this tool
it's not out yet. still in dev. when i get more info, ill announce it on my channel
Fantastic idea. I'll keep a keen eye on its release.
Also, you created a beautiful tune. I couldn't help myself, so I had to import it to Suno with my own AI-modeled vocals and it turned into a beautiful song (with the 80s radio quality of course 😄)
Sadly, I am the only one who can enjoy it.
wow, thanks for sharing
Cakewalk FTW! 😁
Best DAW money can't buy!
🎵 WU HU 🎶
This program seem to lack velocity [dynamics] but has lots of potential.
Perfekt 😃👍
i am music producer 25 years,, and i use this only for inspiration
It's only a matter of time before more and more and more and more AI generated music, including every instrument and every vocal, will start to dominate what people listen to.Hopefully I'm wrong about dominate.
@@bluzedogg Udio already do better vocals
This is amazing.
😃
you can do this with udio or suno . its tricky but you can do it after several tries.
The concept is cool. A true musician wouldn’t need this though. But for home recordings, it can be helpful for lazy producers.
Now if it was real-time, that would be cool, so you can improvise with a band at home in your bedroom.
As a musician, for writing, I like the part of coming up with and writing all the instrument parts, but for improv jamming, it would be awesome (if it could be real-time).
For amateurs, producers who don’t play instruments or musicians who only know one instrument and probably use loops to fill space, it can be helpful.
Speaking of loops, loops pretty much do the same thing but sound way more mastered.
I like the comments about it doing this in midi. So the sound quality wouldn’t be an issue.
So, doing it in midi, in real-time, following you while recording, now you’ve got a winner.
Its possible add voices..?
This is crazy. I'm curious if it could be combined with a tool like MicroMusic, since it turns any sound into a synthesizer preset. If you'd like to try it I can give you a free license / affiliate link
Excellent video, I love that it doesn't replace the artist like udio, but is a tool. The title is not clickbait, this workflow truly looks like the future of music production.
*WHEN... WHEN... WHEN...I WANT TO PLAY TOO... WHEN???... DID I SAY WHEN?????*
so damn nice
Like Output coproducer but much more to the point, at least for my uses. Useful when I have a certain sound in mind but not a sample or instrument that is quite what I want. The guitar strumming with a big better quality could be excellent.