EXTRACT STEMS FROM ANYTHING!!

[@MrBillsTunes]

I wanted to put a few notes here to clear some things up: 1. The m4l device is actually *FREE*, if you read azuki's Bandcamp description you'll see the download link's right there! 2. The device you'll get from azuki's Bandcamp page is for Max For Live only (ie. you need Ableton Suite to run this!) 3. If you're not willing to go to the effort of reading some things + doing some potential installation troubleshooting + messing around with a VM and/or installing Python, FFMPEG & Spleeter.. I would highly suggest not even trying to get this working. I wasn't lying when I said in the video it's somewhat of a pain in the ass to install! You've been warned :p

[@flaviuspopan8024]

Mr. Bill Do you know if her plugin is configured to export stems using the model that supports splits with 11KHz and up? Default Spleeter models only gives you splits up to 11KHz but you can manually specify that on the command line to use up to 16KHz.

[@FuZZbaLLbee]

Flavius Popan I guess all the plugin does is get the selected wave from Ableton save it and call the python script

[@flaviuspopan8024]

@@FuZZbaLLbee Checked the source on the config and it seems to be using the default `spleeter:4stems` model, so all the splits cap out at 11Khz which isn't too bad unless you want better acapella splits. You can tell it to use the models trained up to 16KHz here github.com/deezer/spleeter/wiki/5.-FAQ#why-are-there-no-high-frequencies-in-the-generated-output-files-

[@bcrypt]

@@flaviuspopan8024 11khz for now, though it's easy to modify in the code. main reason being i benchmarked it and 11khz is 30% faster. but if people want i can change it.

[@bcrypt]

if anyone runs into an issue, please file it at github.com/diracdeltas/spleeter4max/issues. (but read the troubleshooting guide first: github.com/diracdeltas/spleeter4max#troubleshooting-and-faqs)

[@Eriktellman]

Came for the plugin, got half a college degree in ableton. Who the f* is this dude he's op

[@xitau2151]

If you wanna learn everything in ableton, Mr bill, Tom cosm and sadowick are your guys

[@WeSLeYxMaNTooTH]

mr bill is a god

[@jz5748]

mr bill been teaching ableton for about a decade...

[@captainadmiral1]

He is the avocado 🥑 frog master 🐸

[@mohitrahaman]

He is Mr. Bill

[@michaelcorass215]

Your method for scrolling a sample at 13:01 made my jaw drop. I haven't touched music in years, mostly because of how tedious it became, all the fun dissipated in that weird way technology does to us. I now have hours of ideas suddenly worth playing with because of 28 seconds of this video. It's so obvious, I'm exasperated thinking about all the time I've wasted. Thank you!

[@nocanhave120]

hold alt/option and drag the first play head.

[@MagnoliaBeats]

I just realised I've been subbed as a 'Hardcore Abletoneer' for longer than any of my relationships. You treat me so good, Bill.

[@llewodcm20]

Thank you, Billiam. No filler, just info. You're a KZbin saint and an Ableton wizard.

[@TribecaBlackstarr]

putting field recordings from shows in here has been cool too......like the audio barrier between smoking pit conversations with a touch of a band jamming in the background has been some of the cooler shit ive got from this

[@rkken]

just found out you can easily scroll through samples with ctrl shift and dragging the waveform. makes chopping way easier than before

[@MrXikwriNeyrra]

The only feature that I used to miss from Cubase now on Ableton! 😎

[@clericsupport]

Came here to comment this

[@noakuu393]

Is this only in Ableton 10? I can't get any of these features everyone talks about in 9

[@hanskracht3528]

@@noakuu393 yeah 10 only

[@matrixate]

Great to see small guys getting busy with this tech. Just for those of us that have RX...this is basically doing the exact same thing as RX Music Rebalance. I don't actually use it but for remixing and resampling tracks, it is totally essential. Thanks for sharing.

[@cameronhawkins4877]

I held off on watching this for a long time because i was positive it was clickbait. This is mind-blowing

[@nothingbutdope5316]

same here. sad, right?

[@Melgaer]

just a clarification: Docker is not a VM. It's os level virtualization. It might sound nerdy, but thats quite important: Docker has almost negible impact on your computer performance.

[@geronimocappelli790]

Hey Bill! Around the 13:00 mark you talk about scrolling around an audio sample. You can also quickly do that by Alt-clicking and dragging the arrow playhead on the clip! (no need for the deactivated placeholder clip)

[@MrBillsTunes]

Geronimo Cappelli yeah I know it just feels weird to me to do it that way for some reason 😅

[@AlfredNazGates]

I picked up glitch hop and producing with mr bill, took some years off, now I'm back and hes still at it. Master Bill.

[@thehitter2708]

Bro you're a very good teacher and super thorough with your details. Makes everything easy to follow and I learned a few really cool workflow tips

[@marscrasher]

there truly is no such thing as writers block. just watch a mr bill vid

[@icethink6280]

duuude am going tru this rn!! Thanks Bill!!

[@jonaswang3936]

Good one

[@nicoschmitt9857]

for the first time i'm happy i did this apprenticeship in IT xD installation was a cakewalk

[@ify105]

This is absolutely game changing

[@7minutesdead]

This is crazy useful. I really love to start ideas from other songs. This makes it a lot less limiting if it splits out things like drums and vocals away from the rest. A lot more stuff usable for foundations and inspiration. Thank you for showing this

[@IRONSILVER22]

You can also do this with Acon Digital Acoustica, it's more expensive, but has a GUI interface and easier installation. You can also tell it what to look for, which improves results. This is a super cool tool though, especially at this pricepoint.

[@TribecaBlackstarr]

deosnt even sound as good! each sound digital as fuck at the end of the day, unless its super super clean source, maybe one of the tracks will sound 'great'.....typically spits out digital as fuck audio regardless what is it $50+ for the acon? free for the alt

[@midnightmartymusic]

@@TribecaBlackstarr i just learned about acon acoustica but haven't had a chance to play around with it yet (assuming there is free trial). But I am curious, are you saying spleeter and acon (hell, throw izotope rx7 into the mix) typically end up sound super digital? Is there a favorite of yours?

[@killianmerdrignac2077]

How freaking revolutionary is this! Think I'd use this for just how incredibly useful you can sample songs and use stems for production or even live purposed. Wow 😱 Thanks man!

[@MrBillsTunes]

If you're having troubles with anything azuki has made a Discord server where you can interact directly & ask questions: discord.gg/zKzYR4D

[@rosesprog1722]

Or go to EZSTEMS.COM

[@MrBillsTunes]

Yep, or go to ezstems which takes longer, doesn't do real-time processing (like azuki's new VST3 does) & causes you to have to tab out of your DAW (probably fucking up your flow-state). Can't say I'd agree that's a better alternative!

[@Lalaland.001]

@@MrBillsTunes don't know man, Ezstems and Melody basically do the same thing without eating away at your cpu capacity. Also the time you need to wait for spleeter to do the stems is the same time maybe a bit longer to do it online. So either way your flowstate gets fucked up. Plus side of the online AI that it gets used a lot more, more data input means better data output, hence getting way better then an in-app or even worse in a virtual machine. Also you don't need Ableton suite or max for live, and it does the exact same thing for no extra money. I can't say I agree with your statement entirely. It does look cool though. But it still doesn't justify the tedious installation, wasting time on it or even remotely putting forth the effort of even wanting to do it. And spleeter is already fucking up my workflow just thinking about having to jump through those hoops. I think I'll pass...Just my 2 cents, sorry not sorry...

[@juanchis.investigadorsonoro]

@@Lalaland.001 yeah... 1 dollar is a lot of money to put into it.

[@AlmostChadSmith]

@@juanchis.investigadorsonoro 😂

[@CreativeRamy]

Omg this thing is nuts! I wish installation was a little more easier though, or maybe you can do a video on how you installed it for us newbies? (I'm an Ableton newbie) I loved how you showed the use cases too to widen our production scope of thinking great work man!! Subbed so I can possibly learn more from you man!

[@obber2]

I am very glad to not be the only one using Resonic Pro as my media player

[@beatz04]

Learned so much more from this single tutorial, enough for like a month to try out new avenues. Thanks !

[@hectorsaucedo4465]

It's super easy. Honestly I watched this video azukis video and like one video that told me that inorder to run docker I needed to turn virtualization on, and then another video that helped get into my bios. Then I just followed this video and azukis video andddd the instructions and it works. And that's coming from someone who doesn't like to put effort into anything. Put a little work now and getting vocals or drums or whatever from a track will only be easier with this.

[@mattryan6886]

This is freaking amazing tool!!! This dude stay dropping Ableton knowledge & fire 🔥 tracks. Mr. Bill DOPE AF

[@GerrySenco]

This changes the game completely

[@danluba]

Dude, the sound-staging on your podcast track is first class.

[@danluba]

Hearing track 2, I can see that stereo sound is not wasted on you.

[@h.n.r_funi3324]

I've always worked with splitter extracting vocals from local songs that i love that i otherwise couldn't get the actual stems for in any other way for me to remix them, i've been doing it since 2016 but the thing is, the vocals have so many artifacts mixed in that some you can find a way to get rid of and some other you won't be able to no matter what you try, i've spent very very long exhausting sessions just trying to process and mix these type of vocals in my remixes that whenever i actually get to work on a studio clean vocal stem every now and then i feel a huge burden lift off of me and would finish the song 1000 percent faster

[@GerbensBackingtracks]

To install this you only need to follow the guide on the github page of spleeter. You only need Python and miniconda, and you only have to put in one command in miniconda to get it installed (which is on the github page) so idk what you mean by you need ffmpeg, as it comes with spleeter.

[@direhavok8732]

Awesome Sauce.... You weren't kidding that this was a hard install. It took me an hour & a half to do so, but finally got Max4Live and VST3 versions running. not sure which I prefer.

[@cborkowski]

FF mpeg is a very old video processing library, it started out as a server side lib to process and convert video files but expanded into desktop variants

[@iDigsGiantRobots]

dude you've taught me so much just in between the points you were making! nice!

[@apoplexiamusic]

that is crazy!!!!!!! what we could only dream of 15 or 20 years ago is now possible!

[@airfixx_8952]

OMG... Mind is blown..... Acapellas of anything now!!!!!!

[@marshmcdan]

This is so awesome, I can't believe no one's developed a AU/VST. I'd buy that for more than $1

[@bleach.princess]

Bill you looking good wtf. Seems like you've been spending your quarantine time right.

[@MrBillsTunes]

Thx, yeah I have this app called "Seconds" on my phone. I've been walking to the park a few x a week & doing some tabata workouts & carrying dumbbells down there some days & lifting & shit... Also been doing 16:8 intermittent fasting (very leisurely) for a while now & I've found that generally helps me feel/look a bit better.

[@satisflomusic]

@@MrBillsTunes Helps get the creative juices flowing! Can I ask what that magical audio player from within file explorer is?

[@satisflomusic]

Nvm just noticed resonic pro..duh

[@bobfrode]

its the hair....hehe Looking good there MrBill :)

[@bcrypt]

@@bobfrode dude bill nice job not mentioning that i gave u a haircut

[@Highcastle_of_Tone]

I hope we eventually get a vst/au plugin of this technology. I'd be happy to pay for this functionality. I'm not really interested in using the stems in my music production, but they'd be great for learning individual parts of songs.

[@MDJ5]

use case #5 - i got spleeeter going on my buddy's computer so he could remove vocals from songs for his karaoke biz back in december. not much use now...

[@jaffe]

had to check out the explorer extension you were using - Resonic Pro - downloaded the player and its already amazing.

[@jaffe]

(it's not an explorer extension for those of you who were curious - it's a stand-alone product. )

[@Reversed82]

eh, the github installation instructions are fine to install it manually, plus you can use the different pre-trained models because there's more than just the 4-out one. (IMO the 2-out one sounds best)

[@TribecaBlackstarr]

depends on the source, ive found..... worth checking the various possible outputs.....usually the bass stem is worthless imo

[@OfficialAHEE]

I can only imagine what copyright lawyers think of this. Lol. Great vid Bill! And big ups Azuki!

[@TribecaBlackstarr]

lol gotta get it in before big brother starts clamping on us......*now exporting 491726 stems*

[@tou-send4349]

me in 2018: “my friends said they want to start rapping and my focusrite came with a free demo of ableton, guess i‘ll try making some beats“ me in 2020: “yeah just need to install python so i can rip this hihat from my fav black midi track“

[@whispy8355]

you need max for live and im 90% sure that it comes with only live suit \

[@tou-send4349]

Whispy i know, i own live standard now (still no ripping for me)

[@gordoncabaniss]

@@whispy8355 you can buy max for live seperately. And run it with Standard.

[@FuZZbaLLbee]

I guess the MAX for life is only needed if you wat to be able within Ableton. The python script will more the likely be able to do this outside of Ableton.

[@kavajarrace]

tou-send black midi already has stems doe

[@slals]

Funny fact (I find it funny, or impressive), Azuki who made this plugin is working on a web browser called Brave.

[@Essenger]

i've been using PhonicMind for this for a few years, curious to test this out and see how it compares

[@AerocityMusic]

Essenger essenger

[@ColorfulBeats]

Let us know about your findings If you have the time!

[@6THSTREETMusic]

This is SO GOOD. Thanks for sharing Mr. Bill!!!!

[@valiumdupeuple]

New quickest way to scroll audio in a clip: hover over a clip, hold shift+alt (+cmd if you want to disable the grid momentarily). The mouse turn into a mickey mouse hand, scroll the audio. (this is the mac shortcut, do the math for Win).

[@omniplexmusic]

Woah that's crazy / incredibly convenient! I'm definitely gonna go the python/ audio library installation route rather than an entire virtual machine

[@chachaloo]

I tried and regretted it. my computer couldn't handle the installation and I found the VM path to be way more convenient. you just open the vm when you need Spleeter and close it afterwords, very simple actually

[@chachaloo]

Ezmyrelda Andrade ok

[@basedvato]

Izotope RX7 does this already to isolate vocals, instrument, subtract drums etc.

[@thefamousdjx]

I've tested various tracks with izotope and it isnt as good as spleeter, not to mention spleeter is free

[@musicart9257]

The best fucking introduction I've seen so far.

[@koredeaderele1666]

very dope. need more libraries like this

[@Bartetmedia]

This will be excellent for my Remix of an old disco song that I'm working on. Thanks!

[@ryancoll22]

Can definitely see DJ's making use of this for mashups and song transitions

[@fredlenox9772]

Great explanation! It makes everything more clear. Except the IT part: Docker is not a VM. Further everything looks great.

[@MikeFlex]

omg that red arrow 🤣 but really this device is insanely cool, thank you for breaking it down for us!

[@valzugg]

holy shit you got me inspired with all these possibilities

[@kyju7093]

OH MY GOD, i dont know what blew my mind more, Spleeter or the fact I somehow wasn’t subscribed to Mr. Bill

[@jcmeyer8979]

For me it was his new haircut

[@battledj2933]

Izotope RX 9 has a similar feature called Music Rebalance. The separation in Izoptope is to my opinion better than Spleeter, less siblings and better separation, but comes with a price of course. :-)

[@MaxSonicUA]

This plugin is mind-blowing🤯

[@sionnachs_workshop]

Bruh you don't have to do the old way to scrub through the audio clips. One of the recent updates for live introduced ctrl + alt + click and drag directly on the audio clip. The mouse pointer has to be half way down the clip I think. Pro tip: you can then ctrl+4 to turn off the grid real quick and then you can scrub yourself into happy, off-time, accidents!

[@martintomorad9164]

Alright, installing Python and required libraries is actually quite simple, here is a quick guide on how to do this on Windows: 1. Download Python 3.7x installer from official Python website. Make sure it's any 3.7 version, Spleeter doesn't work with older and newer versions. 2. Run installer, make sure to check install for all users and add to PATH checkmarks. After that is done, you might have to restart Windows. Now we can simply use pip, a package manager that comes with Python, to install Spleeter and ffmpeg. Here's how to do that: 3. Open cmd or Powershell, make sure you run them as admin(easymode: press Win+R>type in cmd>press ctrl+shift+enter) 4. In cmd(or powershell), type in: "pip install ffmpeg"(without quotation marks). This will install ffmpeg. 5. After that is done, type in: "pip install spleeter"(without quotation marks). That's it, you got everything you need to run this m4l device. Have fun.

[@JonathanOgdenMusic]

So good! Learnt a lot from this, thanks!

[@abdali5035]

Love your new look Mr Bill

[@MrBillsTunes]

Haha you can thank azuki for that also!

[@lachlanjames9320]

Great tutorial as usual, Andrew!

[@peterelfman]

You can blame Andrew Huang and the 4 Producer/1 Sample video for my presence here. Magix has released Stem Maker which does what Spleeter does, but not nearly as well. It's far less complicated to install, but I don't know if it's proprietary to their family of DAW's or if it's available for Ableton. It might be worth looking into, for the sake of simplifying the tech aspect of having a stem-making tool. I wish I could install Spleeter, honestly, it's output is vastly superior.

[@SoundsGoodRecs]

Even faster way to scroll through a sample, hold cmnd + Shift while hovering over the sample when mouse icon changes click and drag left or right. Much quicker.

[@neonmasterva]

Your workflow is so insanely fast, damn

[@onebluebass]

Right?

[@riccardowanderlingh]

This tutorial is bellissimo!!! Mr Bill MITICO!!!!

[@hyperconfidence650]

Regroover by Accusonus does this pretty well too. Can split into 6 elements

[@DaCashRap]

Is it just me or is this thing a gamechanger? Especially for sampling beatmakers.

[@TokoXIII]

We love you Mr. Bill

[@infinaneek]

Ah sweet! It’s like regroover / izotope music rebalance.

[@au5music]

Mind Blown.

[@nissimtrifonov5314]

This intro is genius!

[@brianxaos5855]

this is incredible!! thanks Bill!!

[@Captyvate]

I've never used Ableton but this is dope AF!

[@QauntumCellist]

Lol, Mr. Bill! You're editing is {fire emoji} my dude

[@markl7520]

nice video dude, I like your creativity, good job

[@M4NC3R_420]

cool your doing tutorials man. I love watching you make stuff on twitch too.

[@steveduffy5137]

This has always been the holy grail for us. Can’t believe it’s finally here. Does it provide consistent results across genre or is it really depending on suitable source material?

[@HarmonicaMustang]

It blows my mind that this thing is written in Python. The biggest thing I've written thus far is a random password generator. Granted it's just a compatibility layer to run the TensorFlow machine learning software, but damn. I knew Python was powerful, but this is insane.

[@doseddubs9842]

This works way better than I was expecting. I hope i can get this to workk!

[@cb98678]

So a heads-up for allyou traktor users. I've been able to successfully create a script that feeds my regular MP3 folder into spleeter then takes the output and recombines it into a working MP4. Stem file. Using ffmpeg and bento You can feed your entire library into the script and it takes a about 3 minutes per song. I found this extremely useful for transferring my entire library to working stem files and tractor. Oh the possibilities for remixes are endless

[@nathanaelguirguis7886]

Cant wait to try this out!

[@nomoregoodlife1255]

the power of open source is endless!

[@KarlGilbertOfficial]

can we get a mac installation guide for this awesomeness!?

[@ourfriendgus]

How many times can I upvote per view, so good.

[@AntandraMusic]

Holy guacamole! That's so awesome.

[@gepmrk]

I had a band a few years ago the keyboard player in which used to play with Ian Carr.

[@adamsmith7058]

Nucleus were a badass band back in the day.

[@evandevries1175]

Thanks, Mr. Bill!!!

[@SchultiTube]

Thats crazy! If this is possible, one could make a spectral remover out of it? Like isotope RX?

[@shpongled587]

That track is soooooo good though! 10:30

[@snorresyvertsenhanisch2284]

I have ADHD now Mr Bill, I am to excited for this

[@DarkMetaOFFICIAL]

That's sort of how I feel about money too. I already have some, but it's always kind of nice to have some more 😂

[@chimneychap5142]

It's beautiful how free and open source software allows someone to take bleeding edge AI algorithms and apply it to an Ableton plugin to share with everyone freely. I would normally expect a company to keep this knowledge secret for the sole purpose of taxing people for the advancement of technology. I get music software is a capitalist market aimed at professionals but in this day and age the proper tools should be in everyone's hands if most computers are capable.

[@MDHaughton]

Open Source is the future of software. Back to it's roots, so to speak.

[@dh.l7499]

This is soooo good! Thank you so much for sharing!

[@lostrobotmusic]

I'm an audio plugin developer. If you asked me about this technology an hour ago, I would have said that it's probably impossible, but if it is possible, we have at least another thousand years before we get remotely close to achieving this technology. It is impossible. Yet here we are. Trust me when I say, the more you know about how audio works, the more confused you'll be when trying to figure out how this thing can exist...

[@chrispetersen3613]

Also the more you know about AI and neural networks the more confused you will be about how audio works.

[@lostboyjay]

@@chrispetersen3613 don't know why but i laughed out loud at this interaction

[@BuckingFuffalos69]

If you think this is impossible then man you are gonna flip when you find out the technology they’re hiding from us. It’s essentially magic

[@yokedmusic3867]

Is it just me or does 8:39 sound strangely familiar

[@AdriatixMusic]

dat pornhub intro lmao

[@unfinishedmonkeyrecords]

wow gamechanger... you know of any other plugin using this technique that works outside ableton?

[@jfoldger1184]

@10:31 nice melody bro, cream action

[@zandie12]

Bill low-key lookin pretty fine tho 😍