Spectralayers 10 - extract lead vocal without back vocals

Hello, can I extract only the main vocal from the finished song without backing vocals. I would like to have the same song but without the lead vocal, but to keep the backing vocals in the recording. Is it possible and if so, how?

greetings Dalibor

This would be complex and unlikely to work 100%.

You could try:

  • Extract the entire vocal
  • Use the multichannel Unmix to remove signal from l/r / (This is presuming that the backing vocals are panned and the lead vocal is central)
  • Or home in on centre signal within your DAW using various methods

Otherwise I think this will be almost impossible.

Of course, others may have a better idea.

Not true. If it’s audible then it is possible. I have (personally) unmixed an audio with 3 lead vocals (all panned in the center) along with background vocals (for example a record with someone rapping with a sample playing throughout the record). Of course (the unfortunate thing is) I cannot demonstrate these examples because of copyright issues.

Yes you can.

Without you giving an example it’s hard to explain EXACTLY how, because there’s so many variations and scenarios when it comes to audio. However I can refer you HERE (tap here) where you can see and observe what I was able to do with spectralayers. Also there is another mini tutorial on my tiktok demonstrating how to unmix vocals (whether they are singing in triplets or quarters of unison harmony).

Thank you for your reply. I tried multichannel unmix, but the result is bad, or rather none. An example is a duet, two similar vocals. If you want, I can send you a vocal so you can try it out. I am a regular user of Spectralayers 10. I can send you a wav via e-mail.

1 Like

If you’re going to simply say “Not true”. Then please explain how to do this.

3 lead vocals. (By definition mean singing). Panned central.

By the way you forgot to quote the rest of my post. You know, like the part where I said, “ Of course, others may have a better idea.”

But not to worry.

“Almost impossible” doesn’t mean “impossible”. But even so…

Phil thanks. You are right. I would like that “Unmixing” to give us an explanation of how he managed to separate the 3 vocals that were together in the center. It looks like he did it on a vocal trio where all 3 vocalists were in the center.

1 Like

Please explain how we can do it with Spectralayers 10. I can send you a file.

1 Like

3 lead vocals. All panned mono? All in a mixed song to start with. I would love to know the process. Please enlighten us.

1 Like

Btw your Tic Tok (Maytree) example is an A cappela to start with.

1 Like

@KARUBA @Phil_Pendlebury

I’ve been trying to do this for the past year. Unfortunately it seems like people (like yourself) are unaware of this thing called copyright. Its almost like people are OBLIVIOUS to this idea of copyright and it seems like no one has ever heard of a copyright violation. I would like to demonstrate further (by doing tutorials) how I was able to do stuff like that (and tbh, I have actually have other examples that is much more impressive than the ones I already have on tiktok), unfortunately the examples that I have are in violation of copyrights. For example, I collaborated with another tiktoker and unbeknownst to me the original content creator did not authorize for their copyrights to be used so they removed my examples (which you can view here).

I would like to do tutorials to further demonstrate techniques I use to unmix vocals and sources, unfortunately Steinberg has a strict copyright policy that doesn’t allow me to do that on this forum. Maybe if Steinberg collaborates with other musicians and figure out a way where they can use other musicians music thatvcan be used to demonstrate software (such as Spectralayers) than I would be more than happy to explain further how I was able to do it. I actually want to help other members and my intentions of being on this forum is to help members here.

Other than that, all I can really say (which I’ve been saying from the very beginning, all you have to do is just go back and read my comments) is that the magic wand tool is extremely powerful and it allows you to select huge partials(chunks) without painstakingly selecting each frequency tediously. I find that the sweet spot (for the settings of the magic wand tool) is to set the tolerance to 3.0 db and the maximum width to 4.00 s. I’ve also mentioned numerous times that cutting to layer below (which btw I did demonstrate in that tiktok video. If you observe carefully, I purposefully left a little bit of another vocal in a layer to demonstrate to cut to layer below. Go back and rewatch the video and you will see how I used the cut to layer below to move what sounds right to the correct layers) is how I managed to move audio to the right places (in a non-destructive fashion). Also (unlike the one who does the youtube tutorials demonstrating Spectralayers), I use the shortcuts and the shortcuts allows me to work much more efficiently in a timely fashion. For example, (if I use the magic wand tool) I can select huge chunks of partials by simply selecting(and holding) shift rather than selecting each partial individually.

Also the Casting feature within Spectralayers is extremely underated and undervalued. Casting allows me to carve out frequencies and use those carved frequencies to shape out the layer (for a much cleaner audio layer). For example, lets say you unmix a vocal but you noticed that there is significant bleed from other layers into the vocal layer, you can use the unmix levels feature and unmix the vocal into a high level and a lower level then use the high level to carve out the other frequencies (by merging the other layers together and duplicating the untouched undestructive layer and then phase inverting and merging the two layers and then adding the phase inverted layer back to the high level).

1 Like

Wow that’s a lot of text that says very little. Sorry but it’s easy to put it in words. Showing it actually working is a completely different thing.

Copyright: I’m fully aware. I have a YT channel. (And I do official streams for Steinberg). I can not easily use my own published music and movie work, (of which there’s a lot), as demos. It’s tough you’re right, but there are always legal ways around it, if you need to show demos. For example you can Whitelist certain items (your compositions).

Of course if you’re simply using other people’s music and not making your own then that’s different.

But there’s plenty of copyright free music available to use as demos.

I would say, please try not to assume without checking a bit of background… (See my info). :slight_smile: And please try not to misquote people who are trying to help.

Perhaps you can take up the challenge then and do a small example thing for the OP. I’m sure he’d be very happy.

Anyway, enough from me on this thread.

My very best. P.

@KARUBA please feel free to drop me a pm. Cheers. (I’ll be muting this thread in my feed for now.)


I just explained (that halfway through the video) I used a feature called “cut to layer below”. There was a piece of audio (in that tiktok video) that belonged to another layer, so I “cut to layer below” and that audio merged(was added to that other layer). This is how I did it. It’s not rocket science, there’s no magic formula, there’s no secret button. All I did was re-arranged the bits and pieces of audio to where they belonged. It’s like building legos and figuring out a puzzle, with puzzles you intuitively know what pieces fit where.

I even created a topic (which you can read HERE) that further gives insight into my process and helps you so that you dont make the same mistakes I have made in the past. When I listen back to some of the unmixings I have done (for example some of the ones I did on tiktok) I now hear oddities/artifacts (for example some of the vocalists sounds like they have double echos on them). At the time when I was doing those unmixings I didn’t noticed those artifacts (because the goal was to do those unmixings in 1-2 days in a timely fashion) and what ended up happening is that I became (sort-of) blind/numb to those oddities/artifacts. Sometimes when you play audio over-and-over again (in a short amount of time), your ears starts becoming numb to oddities and artifacts and without realizing it (in a suble manner) your ears gets used to those oddities/artifacts and audio that is wrong (and out of place) starts becoming normal/natural to you (and your ears). That is why I created THIS topic. I created THAT topic to help those who want to do unmixings (like I have done). Basically, it’s a tip on taking breaks so that you can come back to a project with fresh ears and a fresh perspective so that when you playback audio you can INSTINCTIVELY spot and hear oddities/artifacts and immediately act upon your INTUITION on moving audio that is out of place to the right place.

Honestly, if you cannot understand the concept of putting a puzzle together then I wouldn’t know any other way of explaining this to you. I dont know how else to explain all your really doing is putting a puzzle together and figuring out what belongs where. In my first 2 tiktok videos, I simply explained that you’re just putting a puzzle together and figuring out what audio belongs where.

They’re not good examples, for example finding a good copyright free audio of 5 vocalists on a free copyright website is non-existence (believed me I searched and looked). You might find some examples of lead vocals and background vocals on copyright free websites but even those examples are not very good. I like to do hard stuff and work with audio material examples that would be a challenge for me (the goal is to impress people). If you can find a royalty free version of audio where 5-10 different vocalists are singing on a piece of audio, then please post a link here.

I respect everybody (no matter your credentials or how small you are), however when something is not true I will object and when someone is looking for the light and the TRUTH and when someone (like yourself) says there’s no light then I will object and point that person to the light.

If you say [quote=“Phil_Pendlebury, post:2, topic:862481”]
This would be complex and unlikely to work 100%.
[/quote] and then say [quote=“Phil_Pendlebury, post:2, topic:862481”]
I think this will be almost impossible.
[/quote] I’m going to object (no matter the credentials) and point that person to the light and even give examples of myself in the light. It’s not about proving you wrong or belittling your credentials or arguing with you, its about helping others and pointing others to the light and showing others that it is indeed possible.

1 Like

Hey “unmixing”…You’re hiding :slight_smile:

Record (you yourself on Cubase etc) a mixed vocal of you/friends singing lead/harmonies of any old non-copyrighted tune.

Record “Rockin’ Robin” for example…it’s public domain…you and your singer friends can certainly record/sing some 3part tweedly-dees + lead vocal…stack overdubs and lead vocal purposely on top of each other…add a flute too. Make it difficult…record the stacked overdubs in mono.

Or make a song up. Only has to be 20-30 seconds long.

Then…demix it…with your steps shown.

No sr or pa copyright issue.

Personally, I don’t have motivation to create demix tutorials. You seem eager to participate.

Many of us know how to deconstruct the five vocal harmonies onto individual single tracks for say, the chorus of the mono single of Sunday Will Never Be the Same" by Spanky & Our Gang.

Rig up a fake, similar you-yourself-created “thing”. Then demix.

You seem to have the time and are aching to show your skills :slight_smile:

1 Like

By all means, if anyone can find good examples (preferably something like a triplet or quartets of harmonies in unison) of copyright free audio please post links here.

I’m sending example for “Unmixing” 2 voices (2 lines)

Example for you, 2 voices (2 lines)

@Phil_Pendlebury @KARUBA @DosWasBest

This video took a lot of ENERGY from me to make (tap here)

Hopefully it is made plain and clear.

1 Like

“Unmixing” I greatly appreciate your effort and therefore thank you very much. It can be seen that you are an advanced user of the Spectralayer application. To do that, you have to work hard like you did, as I can see in the video, and to do that, I should be an expert like you, but I’m not. As a normal user, I am interested in the routine, one would say “magic button” for separating the vocals in the “a cappella” example. Now I see that it does not yet exist as such. Once again, thank you very much. I think you could be a presenter for Cubase. You could freely present this video on Youtube. They would help many.

As I said before, there are techniques and tips (to work smarter and not harder) to do what you’re trying to do in a more efficient manner. You can manually rebalance the mix so that you can get the noise profile and the overtones to sound as even as possible. As you can see (in the video) you can actually use the left stereo field and cast(carve) out what is in the right stereo filed (and vice-versa the other way around) and convert to mono and phase merge each channel and then convert back into stereo and that should give you clean audio layers. That way you’re killing two birds with one stone by evening out the audio mix while cutting down on time.

If you (or anyone here) feels like this would be time consuming and too much work for you, please do not hesitate to reach out to me. I would be more than happy to assist anyone who wants to separate vocals or instruments (at a fee). I’ve unmixed Orchestral performances, I’ve unmixed Opera performances and project like this usually takes me about 1-2 days (depending on the audio material and length of audio).