I'd really like some help to improve my Spectral Subtraction M4L device. Are you a M4L wizard? Join my Discord server! We're also working on another project which we're calling a Midi Sidechain Shaper discord.gg/7s6HNJQ

16:25 - I like how you played the A / B enough times for us to actually notice something. Sometimes KZfaqrs only show sample data one time each and it isn't enough.

ok this is not what i was looking for, but you are such a funky fresh guy, so i watched the whole video and now i'm gonna clip your vocals.

Really good and interesting video. Thanks for sharing the knowledge!!

Information and video production is 🔥💪🏾

Thanks for sharing such useful information!

Great video Bro! There's so many old rap songs I've been dying to get acapellas of that don't exist. Defintely going to try this out

Really amazing stuff.

dude, your song selections are RAD!!! Can you give us your other recommendations?

Great content, keep it coming

Just in time! Thank you :):)

i literally paused this right before you started talking about using phase inversion with vinyls to test it out with one of my records that has both the instrumental & mixed audio tracks just to come back & find out that i walked right into my own doom 😭

Nice video. I'm using it to show to some university students why it's so hard to extract audio stems. And also how they might want to try and do it. Thanks

I'm getting a good laugh out of your video editing, thanks!

A retired radar tech here. I get what you're saying and it's good. As a tech, my brain works a little differently. Imagine we freeze time and discern one instant of that time. When you heterodyne two frequencies together and plot them on a graph, what you get mathematically is 3 things. The fundamental frequency, And the positive value of the two added together, along with the negative value of the two added together. That is a measurement of modulation. Like you said, if you add the + value and the - value of the one instant, they equal zero and cancel each other out. Now you are left with just the fundamental, which you can still hear in mono. Then you do your second part and lose that. Now, like you said, it only works in phase. Because the + and the - value cancel each other out, only because they equal zero. Now the more out of phase they get, the more those two values, when added together, no longer equal zero when added together. And that increases the more they get out of phase. So there is going to be an artifact leftover. In fact, that is how radar works, it's called MOVING TARGET INDICATION. We only want to see what is moving because airplanes move mountains, don't. We compare the phase. Anything not moving always comes back in phase and cancels out. What is out of phase is moving. The transmitter puts out a 2us pulse as the antenna spins, Each pulse is at a different azimuth. In the receiver, we listen for the pulses to bounce back. When it comes back we strip one freq out, we call the “carrier freq.” Then we take what's leftover, split it, take one path, delay it, and the other path we invert then compare the two together. What is in phase cancels out. What is out of phase does not. Back to music. When signals at any point in time are in phase, they produce, a maximum value which is loudness. Now, we want to cancel out the loudness of certain freq at a certain time. That way, we remove the vocals. I'm wondering if there is an algorithm or function that changes the time of one signal compared to another. I know by changing the time you get the Doppler effect, but not if you listen to the resultant. I would think that if time were slightly adjusted just a few degrees either way on one track compared to another, it would cancel out an unwanted frequency. Right?

15:38 - "feathering" Ooh, that's a good description.

Totally subbed bc of video editing skills

Thank you so much, you are great!!!

This is dope, just used your link.

Flipping love track pant thing 🚀💜🚀💜

Great videos! BIG UPS Bruv,✌️🤟🤘🖖

You just Rock, period !!!!!!!!

Notifications really are a lifesaver =P First 🤙

Really cool

I appreciate you

Help pls🙏🏿 I dowlload the plugin and I followed Installation instructions But it can't to move on specific directed folder, How to install Azuk VST plugin after download it,

Hi, i gave Slynks Spectral Subtractor 1.0 a try today. It worked quite good, with a vinyl version of a 1970s german song, where original and acapella were not quite identical for technical reasons 😛

awesome tutorial man! you are the best. I used till now Izotope Rx to extract vocals and it did a ok job; I suppose tho xtrax would be better

@ 2:50 , does the speaker simply retract back from the forward push? Or does it actually move backwards from its default position?

Can you teach me the fastest way to synchronize vocals with the correct tempo, thank you 😃

Great video! I am trying to get into making my own mixes, after Djing for a few months and getting an understanding of all the softwares, and equipment. Lots of information in here. Very Much appreciate the explanations. I was trying to get Spleeter to work.. but glad I didn't go through it after seeing about x stems.

Slynk Nice job! I need the reverse. I have a VERY old song for which I played everything and sang. I want just the instrumental and don't want my vocals. I presume your device works just fine to accomplish this? Are there any changes I should apply?

I have Adx Trax Pro 3 SP,best "daw" for extract.

what to do when we dont have Instrumental track.... we have only Original Song//// how to get vocals out

Hi Slynk, I have your sample pack (Slynk Presents Ultimate Funky Glitch Hop), which is awesome by the way. I'm just wondering I noticed a few samples in there that wouldn't be royalty free (like the "I've got the power" scratch sample). Is there any way to know which samples are royalty free?

Hey Slynk. Great Video. What's your opinion about RX8?

I have an Imperfect/incomplete method of doing this in FL Studio, using harmor, and stereo phase cancellation. It has 3 flaws 1, Center pan instruments will still be present. Instruments that are even slightly panned, will be quieter at least. 2, Loss of stereo effects, including most importantly, Room Reverb. Can be mitigated with convolution reverb 3, method is incomplete. It's based on layer options in photoshop/gimp for the image resynthesis of 2 different versions of the file (with and without stereo phase cancellation), however the Frequency image, does not currently have a layer mode that functions. the gain image can be processed with the "difference" layer mode. This is because "difference" mode is math by subtracting the RGB of one layer from the other, if they are the same, this reduces to black, and if it is not black, it uses the top layer for that pixel. then you can take the bottom layer, and separately invert and then multiply layer. white (inverted silence) will do nothing to the image, but everything else will darken (make quieter), and that's per pixel, which means volume of a frequency at a time, which is EXACTLY what we want. The issue is the Frequency/Phase image, which does not center around black or white (hex 000000 or FFFFFF) but rather exactly in the middle grey (hex 7F7F7F). Everything above that means the associated partial pitches up, and everything below means the associated partial pitches down. now, obviously, there is some basic math that can be done here, buuut I don't know what exactly it is. Until this step is resolved, this method is largely unusable, because failing at this step, leads to some hot garbage.

Gotta love those fancy MS Paint graphics.

Hi thanks for your video! I have a song and its instrumental,perfectly aligned and with FLAC quality and when I invert the phase I get the acapella, but we can still hear all the instrumental at a very low volume behind. It is not just some effects or noises, it is the whole instrumental. I thought about changing the volume of the instrumental or the original song but nothing works to get rid of it, do you have any ideas? Maybe it is a common problem with an easy solution, but I am a dummie

@slynk you should check out Hit N Mix Infinity for these purposes...

2023, some rap surgery gone wrong brought me here 😂 thanks G 🖤😎🍭

What if you record an "instrumental" with Azuki's plug-in and use that as a phase cancelation?

Hi Slynk, this is a basic question. How do you select a section of the audio track, without removing the selected part? I have learnt how to select the start/end points I am interested of my audio track, then use command + e, but that removes the selected section. Watching your clip around minute 8:17, I see that you select the part you want to place on another track, but that same part, remains in the track you are copying it from. How do you achieve this? Thank you for your help.

Will your Max plug in work in live 9 standard? its telling me its not supported :( Also downloaded the Spleeter DJ but cant properly get ableton to load it.

How does Moises do it then?

It does not work :/ I cannot set it in the master. It does not appear there :/ BTW XTRAX is AWESOME!!! Whoa. I'm mindblown :o

Yues

Couldn't you in theory combine the methods to get a perfect vocal? Like starting out with phase cancelation, and then using something like Azukis vst, or xtrax to clean it up?

@Slynk - You forgot to mention Ezstems (ezstems.com/). Basically it is a free, online version of Azuki's M4L device. Except - it's better: You can upload via KZfaq links OR an audio file... It then spits out all your stems to import into Live. Also, it adds a 5th Stem option, plus bit rate and cut-off options (not that you'd want anything lower than the highest settings...). Free to use, but paid subscribers have cue processing priority. Sometimes there can be a wait... Other times it's super fast. Nevertheless - Awesome video! it's a fascinating topic and your own M4L device looks sweet! Love all of your vids!

Rx8!!! 😁👍🏽

There is a standalone spleeter that works on mac and already bundled w all the python it needs so it's drag n drop into your apps folder. distro by kyab on github

Rx8!

..have you taken a look at izotope rx8?

I Personally think RX8 and Acoustica 7 do a better job but again you would get what you pay for. It is insane tech though and this was a great demo of it.

If I knew this when i was 20, i'd probably be a billionaire right now

I can't dwld the Azuki Spleeter XO

Because you like to eat sandwiches in between your production project I do that and I put a lot of mayonnaise on it. Thank you 🙏

'basically your shit out of luck, and that's why you are here'.... LoL yup ;-)

I’m so tired of the KYLE Beats advertisement

lock sh*t down stems were eventually released for the remix contest

13:42 - I'd call it "artifacty".

Why no one mentions iZotope RX in these videos on KZfaq?!!!! I wish you compared Xtra Stems to iZotope RX at the end :(

While I don't know shit, I suspect that cartopol is changing the results of the fast Fourier transform from Cartesian coordinates to polar coordinates. Poltocar doing the opposite.

Sorry .. can you create a fungki music like otto croy ? But for free :D

Also where is the Funky Fresh Hat Merch?

virtual dj does the same and more. and free

There is no best technic for extracting acapella,only Adx Trac gives very good stereo image result.. Check "acapella service" on bandcamp

am I the only person who found you from the KZfaq audio library?

There is actually an app you can buy that will bridge 32 bit files to be 64 bit files. It’s called Jbridger. It’s a bit of a hassle to acquire and use, but I’ve gotten every 32 bit VST plugin to work using that technique.

And there's iZotope RX8

IF THREE PEOPLE ARE TALKING IN A VIDEO, CAN YOU QUIET TWO OF THE VOICES AND JUST KEEP ONE OF THE VOICES ON THE VIDEO?

You're a sound guy, right? So you know how important the 4 kHz band is in human speech? So _why do you have it practically nulled out on your voice??_ I understand wanting a little extra BASS in the voice, but this just sounds awful.