Always curious to know what AI tools you have been using, music related or otherwise! Please drop a comment to let me know! ...or just say Hi! I always like people saying hi!
Cool video! I normally use Melodyne Studio for this task. I'm a hip hop producer so use it to extract things from samples to create my own elements for tracks. Have you used Melodyne for this? If so how does it compare? Thanks for sharing.
@@SoundSignals You know, I only just started playing around with Melodyne (I used it for vocals on this video: th-cam.com/video/bAzsRzQfirI/w-d-xo.html), and I'm just using the "basic" version, which I don't think has Audio to Midi. Have you tried lalal.ai to extract stems and then Audio to Midi from there? I've done that a few times to great success. Generally, I've found that most Audio to Midi tools work a lot better with less information. When you feed them full tracks, they get confused pretty quickly!
@@TheoreticallyMedia Yes i have used lalala stems for audio midi and it works great! And you are right the less info the better the audio to midi. You might want to demo the Editor version of Melodyne for your own tests, i'd been interested to see how it fares against spotify and also Ableton. In my own test using Melodyne Studio vs Spotify, Spotify actually did a better job, the notes were there with Melodyne, but the timing was off. Spotify was usable straight away, not too much editing, very, very, impressive!
I might jump up to assistant soon, I don’t know if you saw it, but I just released a Punk Rock Morgan Freeman song, where I used an AI voice generator and then ran it through Melodyne. I’m still kinda “learning on the job” with Melodyne- but I can see how in a few more songs, I’ll be needed the additional features!
@@TheoreticallyMedia Tbh i bought Melodyne Studio a few years ago 2nd hand, pretty cheap, just to do audio to midi. At the time the only other decent option was Ableton Live but that was twice the price for the full Suite compared to the Melodyne license. Anyway, even though i bought for that task and didn't think i'd use it beyond that, but i eventually dug into it, learn it and use it on vocals, which it is great for, sorting out the timing and pitch of instruments and figuring out notes, keys, chords etc. It has become an indespensible tool in my box, one of my best investments, can't recommend it enough. Just checked the Morgan song, it's brilliant! And yeah i could see how Melodyne could help for sure. Groove 3 has a very good set of tutorials on how to understand/learn Melodyne 5, that's what i used to learn it.
It's training itself so that it can recognize us using samples and deal with copyright issues. The tech is dope though however artists who want to "make sure they aren't being stolen from" are gonna end up shooting themselves in the foot because they happen to record something that got used before. Imagine a chord progression being copyrighted - not happening, but people do copyright melodies. The problem is that there are only so many melodies. If we look at the classical composers they built on top of what came before and didn't have to deal with copyright strikes on TH-cam. Now we have that problem and we're gonna have more. Lawyers will get rich and right when the tools are cheaper than ever, uneducated artists will make themselves poorer.
sorry-- took a minute to get back to you, got buried under a few other things. Yeah, you raise a really interesting point. There was that fairly famous Coldplay vs Joe Satriani case, which did end up settling, so we don't know the details on how that played out. And there was another one involving Katy Perry. Now that AI is involved and can generate thousands of melodies and songs in the blink of an eye, things are going to get pretty crazy. I know there are some safeguards being propped up. Recently a writer tried to copyright a comic book that they wrote, but an AI "drew"-- and while it was initially slapped down by the Copyright office, as apparently an AI can't hold a copyright, the writer eventually won, as you can copyright AI work, as long as you can prove there was "significant human input. Whatever that means. In the end, I can't help but think of how the old saying "May you live in interesting times" is actually a curse. 😄
@@TheoreticallyMedia the thing is that since we learned to make everything breakdown to 1s and 0s a long time ago, and every CD can be broken down to binary, that means every CD possible could be broken down to 1s and 0s so all CDs possible (and songs, pictures, etc) are just a matter of counting up from 0. Copyright is holding on by a thread and eventually intellectual property will have to go out the window OR we end up with a corporation with the money to pay for lawyers owning everything and then nobody can make a picture, write a song, or even make a movie.
it would suck if this tool was really abused like that by spotify. and i can even imagine a scenario like that. but copyright issues, like issues with copyright itself, are rising more and more nowadays. a lot of people realize that it can't go on like that anymore. So it's only a matter of time until governments are forced to come up with a new solution where poor artists are not fucked over by big companies for dumb reasons anymore. Art must be freed from greed. Art must be freed from people trying to claim ownership on it. Art wants to inspire people and art wants to be referenced and sampled by people. That is the nature of art and the nature of the way humans perceive and use art. Copyright, as we know it, is a perversion.
@@Beatsbasteln Agreed on the hope this isn't abused, but I'll say (somewhat critically) that YT is much more bullish on that front than Spotify. I've actually seen one dude rack up 1 million listens on Spotify by repackaging Lofi beats from other artists. Dude even goes as far as listing the original artist as a "collaborator"-- Copyright, in its base form, is a great concept and one that is meant to protect the original author-- It's just gotten so convoluted by corporate interest to maintain their assets, and honestly, I think chokes creativity. I still need to swing back on the story, but apparently with the rise of Midjourney and Stable Diffusion, Disney is seeking to Trademark their "Style"-- meaning, you could be in breach of Trademark if you draw something that even LOOKS like a Disney property. On the music side, I think we're looking at a disruption even bigger than the Napster debacle. Technology has moved faster than industry, and industry is going to break a lot of kneecaps before it gives up the throne.
@@TheoreticallyMedia it's just pathetic how disney acts like just because people can make sorta disney-ish images they'd be a serious thread to the films or merch they make. but they have the money so you better don't mess with them in court. that's what copyright does. it doesn't save individuals' rights, but just the rights of those who can afford to fight for it.
2:20 I believe the reason the a.i. or what ever audio to midi converter you are using struggles to convert and adds a swing or switching of the notes is most likely due to phase cancellation and so the a.i. interprets the sounds as it hears it. Phasing fits your description of the conversion perfectly.
Figuring out a workflow right now for music in game development and found this older video of yours. Never realized you had such a background in music till a recent call we were in for an AI Tv thing. Really cool tips, btw watched your video How to Write Music with A.I.and noticed the Machine Jam on the table there, I love that little thing. I picked up everything Native Instruments makes back during Covid when I was trying to go full time into production, finally jumped ship this year to open a studio doing Game Development, short films and music, all with the massive help of AI and people like you who filter through the noise. Keep it up man, maybe jam session in the future, I'm a guitarist too :)
That's pretty cool. I like how it removed the drums before analyzing the audio to generate a MIDI track.However it still looked like a mess and a lot of work to tidy up. I'd personally be much more interested in a tool that just looks at the overall chord of each bar (or whatever time interval) and then outputs long notes that represent that chord, so I can just import it into my DAW and decide myself how I wanna arpegiate the stuff
You should look into Scaler (that's the program I used to generate the initial chord progression)-- That has an Audio to Midi Chord analyzer that does pretty much what you're talking about. Although, to be fair, it does go a bit haywire, since it detects melodies and chords at the same time, and sometimes gives you these super elaborate chords that are clearly not in the original song.
@@TheoreticallyMedia nice. scaler is one of those plugins i always wanted to check out. it's so versatile, even though i watched some videos, i didn't even know it has this feature
@@Beatsbasteln I don't want to oversell that particular feature, but it is pretty cool. I think it works best if you think of it as capturing the vibe of a track, as opposed to a direct transcription. Of all the chord analysis/generators (Capt Chords/Cthulhu) I find Scaler to be the most fun and useful!
@@TheoreticallyMedia thanks for the disclaimer. :) i'm well aware of the limitations of audio to midi software. as a plugin developer i know a thing or 2 about that, even though implemented it polyphonically exceeds my current skill level. i always use melodyne for that. in polyphonic mode it is quite capable. but yeah, also one of the tools where you have to post process the midi a lot.. and i wish it also included the vibrato as mpe pitchbends. then it would be absolutely unstoppable imo
It's free and open source, Spotify have licensed it under a free license. Actually it is running locally in your browser, so Spotify could not (if done correctly) read the file. The reason traditional converters more errors is because it's not possible to magically fix that. It's due to how Fourier Transform works. You can't just get the exact frequency back, it has to sample the average to get the frequencies. This also means notes played at different levels have more problems.
Ah, thank you! You’re clearly more well versed here! I’m curious, the models are getting a lot better (since the time this video was published), have you seen anything in the way of a transformer or LLM that “double checks” the work of an audio to midi converter? It seems to me that a lot of the “junk” in an output has to do with frequency confusion. Like, high notes on a piano get bundled in with a ride cymbal. The converter can’t hear which is which, so it kind of bundles them together. It would almost seem to me that a secondary pass through a LLM might be able to take the midi output and sort those errors out. On the topic of Spotify, I actually finally tried out the AI DJ feature they released awhile back. I actually don’t mind it! I tend to listen to a lot of eclectic music across a lot of genres, and it did a pretty good job of navigating a custom playlist.
"Pedantic" is the literal descriptive term, as where "douchebag" is the commonly used parlance; the way "Ableton" is functionally more appropriate being that "live" has the potential to be misunderstood to mean many other things referentially. Oh, and we all thank you for this long awaited clarification.
Thanks very much for the watch! And, yeah-- you're right, but I rarely hear it used (or use it myself) as the whole name. Although, you raise an interesting point, in that Photoshop has managed to become a verb, but no other Adobe product has. And I never foresee someone saying "Let me Ableton that for you."
I once had a script that said "Using Ableton Live live" and didn't have enough coffee to rejigger it to make sense. "Using Ableton Live" just sounded like I was saying the product...I think I ultimately just said "Using Ableton on stage..."
Does anyone else find it odd that out of all the DAWS available, users of Fruity Loops and Ableton are always so damn insecure? If you constantly feel the need to defend the software you use to make music, you might want to think about switching to a different DAW. Or a different profession. Or hobby. Or life. It's such a weird thing to be insecure about.
Oh, 100% agreed. I think it’s a great way of getting out of the dreaded 8 bar loop. I think in a weird way, if it gives you bad ideas, it’s still helpful, as you now have something to work against. Writing is rewriting and all that!
Zplane has some great tools on this. It's saved me some times with a similar layout but alot more features and generally goes on sale. Anyone else tried it?
Looks like Melodyne but online version, and I don't think it's AI since if it's actually an AI it should recommend a better result than that. Well, its just my opinion, but your video is great thank you!
Seems like it's still having trouble with multi instrument tracks. But using it to convert some guitar you are playing to midi might give you better results than a midi guitar. I imagine that you could use this to record a midi drum part without the need for an expensive electronic kit. Just set up some pots and pans. No midi keyboard controller... a cheap kids organ. We've been looking at getting a breath controller to create more realistic brass sections. Getting a decent one can be pricey (for what it is). Any good a whistling? This might do the trick. Whistle the parts into your DAW. Export each part and upload it to one of these AIs. If it recognises pitch, volume fluctuation, pitch bend, vibrato and slides in your whistling....
The pots and pans idea is brilliant! I’ll have to give that a shot! Yeah, you bring up a good point, as this technology evolves, EVERYTHING becomes a midi controller (albeit not in real time). That’s kind of insane and exciting.
Very true. There is a lot of interesting stuff that has happened since this video was posted, but-- I'm still waiting for the next version of Audio to Midi to come out. Fingers crossed that one is super stellar!
Tech that converts pitch to MIDI has been around since the early 80s however, the real exciting and new AI innovations are within how the audio is tracked and analyzed. Neural networks use processes that are inspired by the human brain's sensory processes, memory, contextual queues, patterns and creative intuition. It truly is a significant leap in computing. From my experience, it's far from perfect however, the speed in which it does the work that would take me a decent amount of time is nearly instantaneous. Even after any clean up that I need to do, I still get the work done faster and net positive efficiency and productivity gains. It really just helps me focus on the aspects of production that I love and spend less time on the tedious tasks that bring me no joy. There's absolutely room for improvement but, the recent rate of advancement has been astonishing and is certainly exponential. It's getting really good.
Apparently it doesn't actually upload any files and you can even open the webpage and then use it offline. I'll test that. You can also delete your recent browsing history after of your worried about it doing something after you get back online. You could a little wrapper for it and never take it online. There's an open source Python script you cold download. Someone will undoubtedly make a plug-in with it.
I believe it does, but might get a bit tricky with the bleed of cymbal crashes and whatnot. Give it a try! You might have to do some cleanup on it afterwards, that’s kind of the case with all of these audio to midi Ai things
@@TheoreticallyMedia I tried it late last night and it detected it as an electric piano 😆 I'm going to play with it. If not I found another program ripx. It's not bad too it has a free trial and if I need it longer the cheapest tier one is $99. I really want to get the midi of a few old songs I recorded to audio way back in the day. I had used guitar pro to make the midi files, but I have no idea what I did with those files. In fact I think Im also missing a few songs that I either never recorded or recorded over by accident
I‘ve been searching for some audio to midi tool for ages as I thought it could turn my mouth singing melodies into midi files,instead of playing it on keys. What do you think?
I think it will, as time goes on. In the meantime you might want to look at Avia-- I did a video about it here: th-cam.com/video/bAzsRzQfirI/w-d-xo.html It won't come out with an EXACT copy of the song you input, but rather one that is inspired by it. I find it pretty good for creating a rough sketch of what you're trying to achieve, and then you can further refine it. Let me know if you try it out!
While all these tools are quite amazing, I can't help but wonder how much more music might be created just by playing your preferred instrument and just hitting record?🤔🎶🎹🎶Play On
I wonder what would happen with the translation if it was better music that had more definition and interest. Try some Bach or maybe Paul McCartney next time.
I don’t know if the model has been downgraded since this video was published, but I used it yesterday, pretty simply 4 chord solo piano, and yeah- it struggled. It was weird. Curious, was your bass sound oscillating at all? I can see that being an issue,
@@TheoreticallyMedia ohh .. no man , it was a bass sample on c2 just played on a -40 percent swing rate ... i guess the model is getting trained in the other direction ! :)
It...can be hit or miss. But I think that's pretty much all Audio to Midi convertors...which, I guess is a good thing for us musicians? Still, I do think the clock is ticking for people that make a living doing transcriptions. Within the next year, AI will be able to parse a whole score, I'm sure.
Always curious to know what AI tools you have been using, music related or otherwise! Please drop a comment to let me know!
...or just say Hi! I always like people saying hi!
Cool video! I normally use Melodyne Studio for this task. I'm a hip hop producer so use it to extract things from samples to create my own elements for tracks. Have you used Melodyne for this? If so how does it compare? Thanks for sharing.
@@SoundSignals You know, I only just started playing around with Melodyne (I used it for vocals on this video: th-cam.com/video/bAzsRzQfirI/w-d-xo.html), and I'm just using the "basic" version, which I don't think has Audio to Midi. Have you tried lalal.ai to extract stems and then Audio to Midi from there? I've done that a few times to great success. Generally, I've found that most Audio to Midi tools work a lot better with less information. When you feed them full tracks, they get confused pretty quickly!
@@TheoreticallyMedia Yes i have used lalala stems for audio midi and it works great! And you are right the less info the better the audio to midi. You might want to demo the Editor version of Melodyne for your own tests, i'd been interested to see how it fares against spotify and also Ableton. In my own test using Melodyne Studio vs Spotify, Spotify actually did a better job, the notes were there with Melodyne, but the timing was off. Spotify was usable straight away, not too much editing, very, very, impressive!
I might jump up to assistant soon, I don’t know if you saw it, but I just released a Punk Rock Morgan Freeman song, where I used an AI voice generator and then ran it through Melodyne.
I’m still kinda “learning on the job” with Melodyne- but I can see how in a few more songs, I’ll be needed the additional features!
@@TheoreticallyMedia Tbh i bought Melodyne Studio a few years ago 2nd hand, pretty cheap, just to do audio to midi. At the time the only other decent option was Ableton Live but that was twice the price for the full Suite compared to the Melodyne license. Anyway, even though i bought for that task and didn't think i'd use it beyond that, but i eventually dug into it, learn it and use it on vocals, which it is great for, sorting out the timing and pitch of instruments and figuring out notes, keys, chords etc. It has become an indespensible tool in my box, one of my best investments, can't recommend it enough.
Just checked the Morgan song, it's brilliant! And yeah i could see how Melodyne could help for sure. Groove 3 has a very good set of tutorials on how to understand/learn Melodyne 5, that's what i used to learn it.
I'm way more impressed for what Scaler can do now, that sounds you made with Scaler is amazing! would be cool a tutorial how you did it.
LOVE Scaler! I think I'm still on 2.7. Have they updated recently? I should check in
I prefer to hum melodies when composing so this is awesome, thanks!
Ha, do you do that thing where you hum into your phone's voice memo? I do that ALL the time!
It's training itself so that it can recognize us using samples and deal with copyright issues. The tech is dope though however artists who want to "make sure they aren't being stolen from" are gonna end up shooting themselves in the foot because they happen to record something that got used before. Imagine a chord progression being copyrighted - not happening, but people do copyright melodies. The problem is that there are only so many melodies. If we look at the classical composers they built on top of what came before and didn't have to deal with copyright strikes on TH-cam. Now we have that problem and we're gonna have more. Lawyers will get rich and right when the tools are cheaper than ever, uneducated artists will make themselves poorer.
sorry-- took a minute to get back to you, got buried under a few other things. Yeah, you raise a really interesting point. There was that fairly famous Coldplay vs Joe Satriani case, which did end up settling, so we don't know the details on how that played out. And there was another one involving Katy Perry. Now that AI is involved and can generate thousands of melodies and songs in the blink of an eye, things are going to get pretty crazy.
I know there are some safeguards being propped up. Recently a writer tried to copyright a comic book that they wrote, but an AI "drew"-- and while it was initially slapped down by the Copyright office, as apparently an AI can't hold a copyright, the writer eventually won, as you can copyright AI work, as long as you can prove there was "significant human input. Whatever that means.
In the end, I can't help but think of how the old saying "May you live in interesting times" is actually a curse. 😄
@@TheoreticallyMedia the thing is that since we learned to make everything breakdown to 1s and 0s a long time ago, and every CD can be broken down to binary, that means every CD possible could be broken down to 1s and 0s so all CDs possible (and songs, pictures, etc) are just a matter of counting up from 0. Copyright is holding on by a thread and eventually intellectual property will have to go out the window OR we end up with a corporation with the money to pay for lawyers owning everything and then nobody can make a picture, write a song, or even make a movie.
it would suck if this tool was really abused like that by spotify. and i can even imagine a scenario like that. but copyright issues, like issues with copyright itself, are rising more and more nowadays. a lot of people realize that it can't go on like that anymore. So it's only a matter of time until governments are forced to come up with a new solution where poor artists are not fucked over by big companies for dumb reasons anymore. Art must be freed from greed. Art must be freed from people trying to claim ownership on it. Art wants to inspire people and art wants to be referenced and sampled by people. That is the nature of art and the nature of the way humans perceive and use art. Copyright, as we know it, is a perversion.
@@Beatsbasteln Agreed on the hope this isn't abused, but I'll say (somewhat critically) that YT is much more bullish on that front than Spotify. I've actually seen one dude rack up 1 million listens on Spotify by repackaging Lofi beats from other artists. Dude even goes as far as listing the original artist as a "collaborator"--
Copyright, in its base form, is a great concept and one that is meant to protect the original author-- It's just gotten so convoluted by corporate interest to maintain their assets, and honestly, I think chokes creativity.
I still need to swing back on the story, but apparently with the rise of Midjourney and Stable Diffusion, Disney is seeking to Trademark their "Style"-- meaning, you could be in breach of Trademark if you draw something that even LOOKS like a Disney property.
On the music side, I think we're looking at a disruption even bigger than the Napster debacle. Technology has moved faster than industry, and industry is going to break a lot of kneecaps before it gives up the throne.
@@TheoreticallyMedia it's just pathetic how disney acts like just because people can make sorta disney-ish images they'd be a serious thread to the films or merch they make. but they have the money so you better don't mess with them in court. that's what copyright does. it doesn't save individuals' rights, but just the rights of those who can afford to fight for it.
2:20 I believe the reason the a.i. or what ever audio to midi converter you are using struggles to convert and adds a swing or switching of the notes is most likely due to phase cancellation and so the a.i. interprets the sounds as it hears it. Phasing fits your description of the conversion perfectly.
Now that makes sense!! Thanks for clearing up that mystery!
The swing is probably caused by the drums that has a bit of swing
My guess is the drum transients interrupts the tones.
@@flwhitehorn You're right
Figuring out a workflow right now for music in game development and found this older video of yours. Never realized you had such a background in music till a recent call we were in for an AI Tv thing. Really cool tips, btw watched your video How to Write Music with A.I.and noticed the Machine Jam on the table there, I love that little thing. I picked up everything Native Instruments makes back during Covid when I was trying to go full time into production, finally jumped ship this year to open a studio doing Game Development, short films and music, all with the massive help of AI and people like you who filter through the noise. Keep it up man, maybe jam session in the future, I'm a guitarist too :)
That's pretty cool. I like how it removed the drums before analyzing the audio to generate a MIDI track.However it still looked like a mess and a lot of work to tidy up. I'd personally be much more interested in a tool that just looks at the overall chord of each bar (or whatever time interval) and then outputs long notes that represent that chord, so I can just import it into my DAW and decide myself how I wanna arpegiate the stuff
You should look into Scaler (that's the program I used to generate the initial chord progression)-- That has an Audio to Midi Chord analyzer that does pretty much what you're talking about. Although, to be fair, it does go a bit haywire, since it detects melodies and chords at the same time, and sometimes gives you these super elaborate chords that are clearly not in the original song.
@@TheoreticallyMedia nice. scaler is one of those plugins i always wanted to check out. it's so versatile, even though i watched some videos, i didn't even know it has this feature
@@Beatsbasteln I don't want to oversell that particular feature, but it is pretty cool. I think it works best if you think of it as capturing the vibe of a track, as opposed to a direct transcription. Of all the chord analysis/generators (Capt Chords/Cthulhu) I find Scaler to be the most fun and useful!
@@TheoreticallyMedia thanks for the disclaimer. :) i'm well aware of the limitations of audio to midi software. as a plugin developer i know a thing or 2 about that, even though implemented it polyphonically exceeds my current skill level. i always use melodyne for that. in polyphonic mode it is quite capable. but yeah, also one of the tools where you have to post process the midi a lot.. and i wish it also included the vibrato as mpe pitchbends. then it would be absolutely unstoppable imo
@@TheoreticallyMedia hi do you have full demo basic pitch
There's also a VST version of it: Neuralnote, which is also free/open source
I'll check it out!
It's free and open source, Spotify have licensed it under a free license. Actually it is running locally in your browser, so Spotify could not (if done correctly) read the file.
The reason traditional converters more errors is because it's not possible to magically fix that. It's due to how Fourier Transform works. You can't just get the exact frequency back, it has to sample the average to get the frequencies. This also means notes played at different levels have more problems.
Ah, thank you! You’re clearly more well versed here!
I’m curious, the models are getting a lot better (since the time this video was published), have you seen anything in the way of a transformer or LLM that “double checks” the work of an audio to midi converter?
It seems to me that a lot of the “junk” in an output has to do with frequency confusion. Like, high notes on a piano get bundled in with a ride cymbal. The converter can’t hear which is which, so it kind of bundles them together.
It would almost seem to me that a secondary pass through a LLM might be able to take the midi output and sort those errors out.
On the topic of Spotify, I actually finally tried out the AI DJ feature they released awhile back. I actually don’t mind it! I tend to listen to a lot of eclectic music across a lot of genres, and it did a pretty good job of navigating a custom playlist.
the name is just a cherry on top :D
Long way too go. Kinda bombed on Vocal Line.
Yeah, I gotta swing back to audio to midi. It’s been awhile since this video and there have for sure been some advancements made!
@@TheoreticallyMediahi, what do you recommend now?
This loop at the start with just the beat and that midi wave
👌🏾 I would dance to that
Haha, like a dummy, I trashed the file after recording this. I never learn the lesson to save everything!
@@TheoreticallyMedia am i allowed to download it as an mp3 n make a beat ?
@@gurudewan3639 oh, 1000%
If you get something you’re happy with, please send it over my way, I’d love to hear it!
THIS WAS interesting, thanks!
Happy to hear! I think a lot of AI MIDI stuff has improved since this video was made, so I’ll be checking back in with it soon!
Ableton is the company name. Live is the product name.
(It would be confusing to say Adobe, when someone meant Photoshop.)
And thank you for this...
"Pedantic" is the literal descriptive term, as where "douchebag" is the commonly used parlance; the way "Ableton" is functionally more appropriate being that "live" has the potential to be misunderstood to mean many other things referentially. Oh, and we all thank you for this long awaited clarification.
Thanks very much for the watch! And, yeah-- you're right, but I rarely hear it used (or use it myself) as the whole name. Although, you raise an interesting point, in that Photoshop has managed to become a verb, but no other Adobe product has. And I never foresee someone saying "Let me Ableton that for you."
I once had a script that said "Using Ableton Live live" and didn't have enough coffee to rejigger it to make sense. "Using Ableton Live" just sounded like I was saying the product...I think I ultimately just said "Using Ableton on stage..."
@@standardqueue 😂😂😂
Does anyone else find it odd that out of all the DAWS available, users of Fruity Loops and Ableton are always so damn insecure?
If you constantly feel the need to defend the software you use to make music, you might want to think about switching to a different DAW.
Or a different profession.
Or hobby.
Or life.
It's such a weird thing to be insecure about.
Seriously been looking for something like this for years, as Ableton is a mess when it comes to audio to midi.
Fantastic! And I still wouldn't call it perfect-- but, at least we have some controls with this one to adjust the output.
@@TheoreticallyMedia I was trying to use it for beat boxing I did, but it isn't for drums, alas.
Love that vibe thanks, new sub
Welcome aboard! Nice pickin' on your channel too!
Every Audio to MIDI converter works MUCH better with just one instrument as an input.
Oh, 100%. But we don’t always have access to stems to use. That said, with this latest crop of audio isolators, even that might become a non-issue!
@@TheoreticallyMedia we can use AI to split the audio tracks and then convert each track to MIDI...
I think if it gets better generating ideas when you're stuck on a loop will be a great tool
Oh, 100% agreed. I think it’s a great way of getting out of the dreaded 8 bar loop. I think in a weird way, if it gives you bad ideas, it’s still helpful, as you now have something to work against. Writing is rewriting and all that!
As long as you keep it a tool..
I am scared that this will stop talent. Even talented people could become lazy.
@@TheoreticallyMedia that’s a cool take 👏🏾
@@MosheBenChaim naw I’ve honestly experimented with a lot of these tools and hard truth is there are no shortcuts
The problem is more the massive amount of ghost notes. You still have hour of work make in suitable for your project
Thank you for the information
How does it compare to Melodyne 5?
And will it be installed on the M1 Mac?
I think this is online only. And Melodyne, honestly, will probably be a lot better. I mean, ha- that is kinda free vs paid!
Zplane has some great tools on this. It's saved me some times with a similar layout but alot more features and generally goes on sale. Anyone else tried it?
I’ll check it out!
THANK YOU SOOOOOOOOOOOO MUCH
Awesome!! You're Welcome!!!
AI can isolate tracks and detect chords so why can’t it midi?
Thanks! That was really useful to me!
Awesome! Really happy to hear that!
what are you playing the beat is so beautiful oh my gosh-
Looks like Melodyne but online version, and I don't think it's AI since if it's actually an AI it should recommend a better result than that. Well, its just my opinion, but your video is great thank you!
Isn't the Midi supposed to split up go back to individual midi instruments and tracks?
You can route the midi to different virtual instruments. Although, that is all done manually.
works nicely with piano audio.
anyways ran it on Firefox, and it cost me 20G of memory on mac.
Has anyone experienced the same issue?
Merci pour l'info :)
Nice tune
Seems like it's still having trouble with multi instrument tracks. But using it to convert some guitar you are playing to midi might give you better results than a midi guitar.
I imagine that you could use this to record a midi drum part without the need for an expensive electronic kit. Just set up some pots and pans.
No midi keyboard controller... a cheap kids organ.
We've been looking at getting a breath controller to create more realistic brass sections. Getting a decent one can be pricey (for what it is). Any good a whistling? This might do the trick. Whistle the parts into your DAW. Export each part and upload it to one of these AIs. If it recognises pitch, volume fluctuation, pitch bend, vibrato and slides in your whistling....
The pots and pans idea is brilliant! I’ll have to give that a shot! Yeah, you bring up a good point, as this technology evolves, EVERYTHING becomes a midi controller (albeit not in real time).
That’s kind of insane and exciting.
thank you bro
100%!!
Thanks for the video
i think the site is down or something. its always "loading files" for me
thanx brother
Glad it helped! I've been doing a lot of AI Image stuff lately, but I do have a circle back to music stuff in the pipeline!
They call it "AI" but this stuff has been around for decades...
Very true. There is a lot of interesting stuff that has happened since this video was posted, but-- I'm still waiting for the next version of Audio to Midi to come out. Fingers crossed that one is super stellar!
Tech that converts pitch to MIDI has been around since the early 80s however, the real exciting and new AI innovations are within how the audio is tracked and analyzed. Neural networks use processes that are inspired by the human brain's sensory processes, memory, contextual queues, patterns and creative intuition. It truly is a significant leap in computing. From my experience, it's far from perfect however, the speed in which it does the work that would take me a decent amount of time is nearly instantaneous. Even after any clean up that I need to do, I still get the work done faster and net positive efficiency and productivity gains. It really just helps me focus on the aspects of production that I love and spend less time on the tedious tasks that bring me no joy. There's absolutely room for improvement but, the recent rate of advancement has been astonishing and is certainly exponential. It's getting really good.
Thanks
Thank you so much!!
Thank you !
real-time vocal audio to midi ? does this work ?
Sorta. It seems to have gotten a little worse over time. Try it out though!
@@TheoreticallyMedia vochlea seems to be a new hot item now. what do you think?
Apparently it doesn't actually upload any files and you can even open the webpage and then use it offline. I'll test that. You can also delete your recent browsing history after of your worried about it doing something after you get back online. You could a little wrapper for it and never take it online. There's an open source Python script you cold download. Someone will undoubtedly make a plug-in with it.
Thank you
Does this work with drums? I have an audio MIDI I recorded years ago and wanting the MIDI of that
I believe it does, but might get a bit tricky with the bleed of cymbal crashes and whatnot. Give it a try! You might have to do some cleanup on it afterwards, that’s kind of the case with all of these audio to midi Ai things
@@TheoreticallyMedia I tried it late last night and it detected it as an electric piano 😆 I'm going to play with it. If not I found another program ripx. It's not bad too it has a free trial and if I need it longer the cheapest tier one is $99. I really want to get the midi of a few old songs I recorded to audio way back in the day. I had used guitar pro to make the midi files, but I have no idea what I did with those files. In fact I think Im also missing a few songs that I either never recorded or recorded over by accident
I‘ve been searching for some audio to midi tool for ages as I thought it could turn my mouth singing melodies into midi files,instead of playing it on keys. What do you think?
I use Imitone
They are always in Beta, so l get the latest developments. 😊😊
doesnt really work out for any time signature than 4/4 but still got the notes right which is cool
I tried to upload a simple 2 minute piano track, and it kept processing forever...
Yeah, I’m. It sure it’s 100% stable anymore. I’ll keep an eye out for some other options!
Wawu that's good am from Uganda I have my song but I have failed how to start it as in voicing could you help me it's a gospel song
Also Neuralnote app seems to be the same.
according to the site "Note: None of your audio is saved or stored. It all stays on your machine."
how to convert stereo polyphonic audio to midi?
i hope they improve this... i droped a mp3 of a song i wanna make in a game and didnt come out right
I think it will, as time goes on. In the meantime you might want to look at Avia-- I did a video about it here:
th-cam.com/video/bAzsRzQfirI/w-d-xo.html
It won't come out with an EXACT copy of the song you input, but rather one that is inspired by it. I find it pretty good for creating a rough sketch of what you're trying to achieve, and then you can further refine it. Let me know if you try it out!
@@TheoreticallyMedia will do thanks
While all these tools are quite amazing, I can't help but wonder how much more music might be created just by playing your preferred instrument and just hitting record?🤔🎶🎹🎶Play On
nice tool！
uploaded a guitar game file. didn't recognize it
Didn't work too well for me. Processed a wav file with piano chords and harmony. Ableton did a much better job.
Thx for your video but the best Audio to MIDI AI is still us ;)
1000% agree!!
Very impressive, pity it doesn't have the ability to ignore vocals.
I wonder what would happen with the translation if it was better music that had more definition and interest. Try some Bach or maybe Paul McCartney next time.
Yo, don't ever disrespect me looking like the white guy in men in black, ya heard?
Audio pitch to midi... Not drums sadly... I WISH
Yeah, I’ve yet to find really outstanding Drums to Midi. Always looking though!!
Tracker in superior drummer by Toontrack.
I've tried it, and it did not do a good job.
It couldn't even transcribe a one bar single bass note properly ,
I don’t know if the model has been downgraded since this video was published, but I used it yesterday, pretty simply 4 chord solo piano, and yeah- it struggled. It was weird.
Curious, was your bass sound oscillating at all? I can see that being an issue,
@@TheoreticallyMedia ohh .. no man , it was a bass sample on c2 just played on a -40 percent swing rate ... i guess the model is getting trained in the other direction ! :)
I tried it was actually quite terrible
It...can be hit or miss. But I think that's pretty much all Audio to Midi convertors...which, I guess is a good thing for us musicians? Still, I do think the clock is ticking for people that make a living doing transcriptions. Within the next year, AI will be able to parse a whole score, I'm sure.
@@TheoreticallyMedia Yeah but It would be sure a time saver if it worked
Convert midi to audio.
Samplab ? ❤
Haven’t checked in with them recently. Are they killing it?
Not really "mighty"- Crashed by a monophonic 8 bar mp3, 112 bpm Trumpet solo by 64 % and 8 Minutes.
Scaler 2 is NOT free
It is not. Sorry, did I imply that it was in this video? If so, I apologize. It IS fairly cheap though, and super worth it!
@@TheoreticallyMedia , this is the video's tittle
The BEST AI Audio to MIDI tool! (And it's Free!)
i prefere spyware off from youtube and you guys arested
uh, what?
Not so good
it sucks
Crap
Thanks