This is a very logical and ENGINEERING sound way of assessing spectral analysis and processing tools. These tools can produce great results, and seem to be developing each year. It's a fascinating area, and since I work with so many other musicians this area comes up often enough. Thanks for working on this.
Brilliant video. I have been using a number of tools for stem separation. I worked on one song with lots of detail in the track on RX10 / Ripx & Spectral Layers 10 pro... and SL came out top in terms of sound quality and minimising the artifacts. Its got this knack of softening- them without losing too much of the quality. Steinberg did goood!
One thing that gets very much overlooked is creating stems from an old mono recording - say from the 1950s or early 1960s. Then creating a much superior stereo mix. This is what i use it for.
I've been a Spectralayers user for a few years now (currently using version 7). This version 10 is very impressive!! With lots of practice you can manually extract the tiny missing pieces of say , the piano. Nice demonstration!!
Holy crap man that spectral layers is amazing man. Holy crap, you just sold an upgrade here from this demo man. Yes man its remarkable, and will get nothing but better over time
Thank you for this. I use RX, but am looking to get Spectralayers when it is on offer as it is (currently) better. I don't want to give you more work, but if you mixed the stems together to a full mix, and then unmixed them, you could determine how effective the unmix is without the confounding factor of unknown processing in the supplied full mix ;)
wow...this was shocking at how well it did! Obvious and expected artifacts of course, but it really did surprise me...still scraping my jaw off of the floor actually hahaha...I actually have a project that I produced years and years ago that another engineer mixed and mastered...now the client wants to revisit and update that project and I'm having trouble getting the original files back from that engineer...your video makes me wonder if I could actually just "unmix" the pre-mastered versions I have here and update them myself...especially if they do actually null when phase inverted from the original! These are indeed exciting times to live in!!!
Spectra Layers is so perfect for me, I use it on my live concert multi track recordings I get them from Dante Virtual capable live consoles, it works much better on cutting the microphone bleeds from the vocal ore other tracks, it makes my Jobe much more easy and quality vise, a big thanks to Steinberg and also to you to covering me up in Cubase.
Thank you very much. Following the video I feel Real detect better frequencies. The example was the piano which Spectre Layer stems doesn't give the full frequencies
There is no doubt that SL10 is a huge improvement on previous versions and is a big step forward, as you have shown. All unmixing programmes have artefacts and less than perfect results, which you expect, but you can do a lot of remixing with this level of quality. One observation I have is that a typical scenario for using this software is for badly mixed or recorded material which only has a stereo version. I think that is where these programmes are put to the test, since they will not be working with optimal material in those cases, and will have a harder time separating instruments and voices. But those are the scenarios where they are most useful. Having done a project of a theatre musical badly recorded to just stereo, with voices too low in the mix, I tried all these programmes and combined results from a few of them. It eventually worked, but took very many hours of trial and error and painstaking work. But I can confidently state that if I had SL10, it would been much easier and quicker. So we should be very grateful for such progress. It may therefore a useful trial if you try some badly mixed or recorded material and seeing how much you can rescue and improve it, rather than clean, well mixed material..
Thanks for your extensive response. That’s a good point that you may need this most to ‘remix’ poorly mixed songs and that separation will be harder on that kind of material. May give that a try. Another use-case may be to make Dolby ATMOS mixes from only a stereo track. That could maybe even be an automated thing 🤓😎.
I am overall impressed by the demonstration. I am interested in software like this to rebalance bad mixes on problem tracks...usually older recordings that have issues and cannot be remixed. Unfortunately, I usually end up with many audible artifacts. So I can make a better mix, but the audio quality of the track is not good. I basically trade one problem for another. The technology is pretty good though.
I love ripping tracks apart and mixing them in Atmos. Once split I then import the drums to Superior Drummer and add what drum kit I like and same with EZBass for the bass guitar. I had been using Izotope rebalance for the unmix but I have to be honest and admit Spectralayers is slightly better, especially drums and bass. Well done Steinberg another winner !!
@@LanewoodStudios Don't forget once you have done the initial song unmix then take the drum layer on its own and split that into snare cymbals and kick. I also tried the noisy speech to extract the vocals but thats not really an improvement from the song unmix vocal layer. AS for multi voice, thats hit and miss but sometimes gives interesting results that can be useful for creating harmonies
I've been trying to use this program (and others) to resurrect a friends 40 year old recordings without much success, mainly due to the poor recording quality of the originals. I've had better luck by tempo mapping the original tape recordings and converting the audio to midi. The most frustrating aspect of these types of programs is the amount of bleed that occurs and the Comb filtering artefacts. I'm probably better off re-recording the tracks I suppose. As per usual, a great tutorial. Thanks.
I've only tried these during the demo periods (don't have $300 for a license atm) - but I definitely noticed a DRAMATIC improvement between SpectraLayers Pro 9 and SpectraLayers Pro 10 in terms of how well different things get unmixed. Hopefully there can be more instrument unmixing options eventually. Actually, I had a thought, it'd be really neat if they had the ability to queue up different actions to be carried out sequentially (if they don't already have that ability) - for example, if I am unmixing tracks, and want to unmix the track, and then unmix those generated stems into components, and not have to nurse SpectraLayers through the process (and do everything manually), but do other things (ESPECIALLY if it is gonna take a while due to the length of audio being unmixed). Wonder if that makes any sense.
Yes the unmixing has improved a lot. As for your suggestion, it is sort of what they implemented for drums. You can unmix a stereo mix to end up with a stereo drum track. And then you can unmix the stereo drum track into its components (snare, kick, cymbals, etc..). It's now a manual two step process though. So yes if they offer this for more initial layers it makes sense to automate that.
@@LanewoodStudios You know ... I was pondering this today, could it be possible to look at the drum unmixing feature as kindof a subset of the component unmixing (or a highly specialized subset, rather)? (does that even make any sense?)
SpectraLayers helps me with my backing tracks. Seriously cuts down my time. I will usually unmix all parts. Then if I want the sound to be more crisp say like on that guitar, I will just track either all or some of that part to clean it up. The piano, keyboard or organ, I will be playing that live.
I’m surprised how well it works… for learning parts, this is what I would use it for and for that I think it’s amazing… I’m sure future versions of this software will be even better but I think for me the difference will be so small…
Its impressive and good enough for a lot of things like splitting out a drum stem for drum triggering. However they need to get closer but I'm not sure if that's possible especially on non-harmonic instruments.
First of all, thanks for the excellent presentation. I can think of a number of situations I might have used this in the past, mainly to rebalance archival material. For my uses, I would not have stripped a mix entirely to stems for rebalancing, but instead used some of the resulting stems to enhance the existing mix. Which brings up the subject of delay. Are the resulting stems automatically delay compensated to allow direct use in my desired method? Manually adjusting a stem or two is no big deal but I am curious about the default delay compensation.
Great demo! Thank you. I was wondering is it possible to manually go into the "other" stems and put the unidentified sounds in the correct stems where possible? The unidentified stems seem to contain a lot of information.
Yes, you can select those sounds manually and then ”cut to layer below”. So the first thing you will do is to place the ”correct layer” right below the ”other” layer and then in the ”other” layer select and shift+x (cut to layer below). Very handy.
SpectraLayers 10 is a definite improvement over version 9, but it still doesn't match the quality of the stems created by DeMix Pro (I have them both). It's particularly noticeable on piano and guitar stems. Bass tracks are also more well-defined in DeMix. And DeMix also separates backing vocals from the lead vocal.
nice demonstration... but why did you use the original mix which didnt null ? why not render a new mix out of cubase using the original stems and use that as the source for demixing in spectral layers ?
How well do you think that Spectralayers version 10 does vs version 9? Do you feel that the upgrade was worth it... can you tell a notable difference between the two versions for stem separation?
Well I only used SpectraLayers 8 so far. As I understand it, 8 and 9 did not differ very much regarding stem separation. Both from my own experience and from what I've seen from other people, version 10 is quite a big change in how well this works. It's never perfect of course, and never will be, but isolating Vocals, Bass and Drums is really amazing depending on what you want to use the stems for of course. Like I mentioned in the video.
Hi am just starting using spectra layers when I open only shows by default vocal it won’t show other instruments how I can change that anyone can explain that appreciate tks
I don’t quite understand what you mean because when you open a stereo mix in SpectraLayers you should just hear the full mix. Maybe start with a video on the basics of SpectraLayers. I have one from a while back: ua-cam.com/video/nUHRBcZ5xhE/v-deo.htmlsi=ADOywn997GXseU22
How come most all your audio on this end has an underlying distortion to it,I wonder if you sent up your audio just a couple db less if it would go away.
This is a very logical and ENGINEERING sound way of assessing spectral analysis and processing tools. These tools can produce great results, and seem to be developing each year. It's a fascinating area, and since I work with so many other musicians this area comes up often enough. Thanks for working on this.
Thansks and you’re very welcome!
Your unmixing demonstrations are the best ones I've seen.
Thank you very much 👍
Brilliant video. I have been using a number of tools for stem separation. I worked on one song with lots of detail in the track on RX10 / Ripx & Spectral Layers 10 pro... and SL came out top in terms of sound quality and minimising the artifacts. Its got this knack of softening- them without losing too much of the quality. Steinberg did goood!
Yes they did 👍
One thing that gets very much overlooked is creating stems from an old mono recording - say from the 1950s or early 1960s. Then creating a much superior stereo mix. This is what i use it for.
Yes thanks for the tip!
this is literally the only reason i care at all about spectral demixing (call me a fossilhead ig)
Fantastic showcase. Well executed, well explained. Sold me on the unmix function too.
Great to hear!
I've been a Spectralayers user for a few years now (currently using version 7). This version 10 is very impressive!! With lots of practice you can manually extract the tiny missing pieces of say , the piano. Nice demonstration!!
Thank you!
Holy crap man that spectral layers is amazing man. Holy crap, you just sold an upgrade here from this demo man. Yes man its remarkable, and will get nothing but better over time
Yes it is!
Thank you for this detailed test of Spectralayers 10; i think i will order the test version, and probably the full version after the test time.
Yes check it out!
Thank you for this. I use RX, but am looking to get Spectralayers when it is on offer as it is (currently) better. I don't want to give you more work, but if you mixed the stems together to a full mix, and then unmixed them, you could determine how effective the unmix is without the confounding factor of unknown processing in the supplied full mix ;)
Yes I thought of that after I made the video already 😉.
wow...this was shocking at how well it did! Obvious and expected artifacts of course, but it really did surprise me...still scraping my jaw off of the floor actually hahaha...I actually have a project that I produced years and years ago that another engineer mixed and mastered...now the client wants to revisit and update that project and I'm having trouble getting the original files back from that engineer...your video makes me wonder if I could actually just "unmix" the pre-mastered versions I have here and update them myself...especially if they do actually null when phase inverted from the original! These are indeed exciting times to live in!!!
Yes me too 😎😁. Give it a try with unmix I would say.
Spectra Layers is so perfect for me, I use it on my live concert multi track recordings I get them from Dante Virtual capable live consoles, it works much better on cutting the microphone bleeds from the vocal ore other tracks, it makes my Jobe much more easy and quality vise, a big thanks to Steinberg and also to you to covering me up in Cubase.
Yes I can imagine that it's great for that job. It's quite amazing what it does. Thanks for your comment on how you use SpectraLayers!
Thank you very much. Following the video I feel Real detect better frequencies. The example was the piano which Spectre Layer stems doesn't give the full frequencies
Thanks for watching and commenting!
There is no doubt that SL10 is a huge improvement on previous versions and is a big step forward, as you have shown. All unmixing programmes have artefacts and less than perfect results, which you expect, but you can do a lot of remixing with this level of quality. One observation I have is that a typical scenario for using this software is for badly mixed or recorded material which only has a stereo version. I think that is where these programmes are put to the test, since they will not be working with optimal material in those cases, and will have a harder time separating instruments and voices. But those are the scenarios where they are most useful.
Having done a project of a theatre musical badly recorded to just stereo, with voices too low in the mix, I tried all these programmes and combined results from a few of them. It eventually worked, but took very many hours of trial and error and painstaking work. But I can confidently state that if I had SL10, it would been much easier and quicker. So we should be very grateful for such progress.
It may therefore a useful trial if you try some badly mixed or recorded material and seeing how much you can rescue and improve it, rather than clean, well mixed material..
Thanks for your extensive response. That’s a good point that you may need this most to ‘remix’ poorly mixed songs and that separation will be harder on that kind of material. May give that a try. Another use-case may be to make Dolby ATMOS mixes from only a stereo track. That could maybe even be an automated thing 🤓😎.
@@LanewoodStudios Yes, good point. That would be a great use of this kind of remixing - making spatial versions of stereo tracks.
@@MACEASY2 exactly! Another video idea ;).
Thanks!
👍
Thank you. Low middle seems to fade to ambiance. Separating live abstract tracks for production reference is my primary usage.⚕️🎶🥀🌀🎹
Thanks for sharing!
I am overall impressed by the demonstration. I am interested in software like this to rebalance bad mixes on problem tracks...usually older recordings that have issues and cannot be remixed. Unfortunately, I usually end up with many audible artifacts. So I can make a better mix, but the audio quality of the track is not good. I basically trade one problem for another. The technology is pretty good though.
Try it with SpectraLayers because the individual layers perfectly sum.
I love ripping tracks apart and mixing them in Atmos. Once split I then import the drums to Superior Drummer and add what drum kit I like and same with EZBass for the bass guitar. I had been using Izotope rebalance for the unmix but I have to be honest and admit Spectralayers is slightly better, especially drums and bass. Well done Steinberg another winner !!
Great tip!
@@LanewoodStudios Don't forget once you have done the initial song unmix then take the drum layer on its own and split that into snare cymbals and kick. I also tried the noisy speech to extract the vocals but thats not really an improvement from the song unmix vocal layer. AS for multi voice, thats hit and miss but sometimes gives interesting results that can be useful for creating harmonies
@@uselessoldman7964 thanks for the tips!
I've been trying to use this program (and others) to resurrect a friends 40 year old recordings without much success, mainly due to the poor recording quality of the originals. I've had better luck by tempo mapping the original tape recordings and converting the audio to midi. The most frustrating aspect of these types of programs is the amount of bleed that occurs and the Comb filtering artefacts. I'm probably better off re-recording the tracks I suppose. As per usual, a great tutorial. Thanks.
Yes if the recording is pretty bad it will all work less well. Probably nothing will work in that case though.
I've only tried these during the demo periods (don't have $300 for a license atm) - but I definitely noticed a DRAMATIC improvement between SpectraLayers Pro 9 and SpectraLayers Pro 10 in terms of how well different things get unmixed.
Hopefully there can be more instrument unmixing options eventually.
Actually, I had a thought, it'd be really neat if they had the ability to queue up different actions to be carried out sequentially (if they don't already have that ability) - for example, if I am unmixing tracks, and want to unmix the track, and then unmix those generated stems into components, and not have to nurse SpectraLayers through the process (and do everything manually), but do other things (ESPECIALLY if it is gonna take a while due to the length of audio being unmixed). Wonder if that makes any sense.
Yes the unmixing has improved a lot. As for your suggestion, it is sort of what they implemented for drums. You can unmix a stereo mix to end up with a stereo drum track. And then you can unmix the stereo drum track into its components (snare, kick, cymbals, etc..). It's now a manual two step process though. So yes if they offer this for more initial layers it makes sense to automate that.
@@LanewoodStudios You know ... I was pondering this today, could it be possible to look at the drum unmixing feature as kindof a subset of the component unmixing (or a highly specialized subset, rather)? (does that even make any sense?)
In my opinion the stems from SpectralLayers are ok for analysing purposes or for rebalancing. They cannot replace real stems. Maybe "not yet".
I agree with you. Real stems are still better, but I am always amazed by how far we got already.
SpectraLayers helps me with my backing tracks. Seriously cuts down my time. I will usually unmix all parts. Then if I want the sound to be more crisp say like on that guitar, I will just track either all or some of that part to clean it up. The piano, keyboard or organ, I will be playing that live.
Good idea 👍
I’m surprised how well it works… for learning parts, this is what I would use it for and for that I think it’s amazing… I’m sure future versions of this software will be even better but I think for me the difference will be so small…
Yes 👍
Its impressive and good enough for a lot of things like splitting out a drum stem for drum triggering. However they need to get closer but I'm not sure if that's possible especially on non-harmonic instruments.
Yes exactly. How far can you go with this?
First of all, thanks for the excellent presentation. I can think of a number of situations I might have used this in the past, mainly to rebalance archival material. For my uses, I would not have stripped a mix entirely to stems for rebalancing, but instead used some of the resulting stems to enhance the existing mix.
Which brings up the subject of delay. Are the resulting stems automatically delay compensated to allow direct use in my desired method? Manually adjusting a stem or two is no big deal but I am curious about the default delay compensation.
Hi, well when I sum the stems and reverse the phase of the original there was silence. So the stems are perfectly aligned with the original.
@@LanewoodStudios Ah, yes, you did show that. Thanks for your reply.
@@artysanmobile 👍
Great demo! Thank you. I was wondering is it possible to manually go into the "other" stems and put the unidentified sounds in the correct stems where possible? The unidentified stems seem to contain a lot of information.
Not that I'm aware of.
Yes, you can select those sounds manually and then ”cut to layer below”. So the first thing you will do is to place the ”correct layer” right below the ”other” layer and then in the ”other” layer select and shift+x (cut to layer below). Very handy.
Great, thanks for the tip!
SpectraLayers 10 is a definite improvement over version 9, but it still doesn't match the quality of the stems created by DeMix Pro (I have them both). It's particularly noticeable on piano and guitar stems. Bass tracks are also more well-defined in DeMix. And DeMix also separates backing vocals from the lead vocal.
Haven’t tried that one yet 😋.
nice demonstration... but why did you use the original mix which didnt null ? why not render a new mix out of cubase using the original stems and use that as the source for demixing in spectral layers ?
Thanks, and yes I thought about that later too 😬.
Afternoon removing the vocal, if I put my own vocal and upload it. Will there be copyright?
Yes, the copyright is on the song!
Excellent.. but can u remove reverb n delay from vocal 👍
Dear Sir.. I expected n answer as there's n option to trigger less reverb..
Thnx.
Yes there is a reverb reduction process in SpectraLayers as well. Have not tried it myself yet though. If I like it, I may make a video about it 😉.
@@LanewoodStudios Thnx Sir..
Cool. But on a laptop...When I listen, it's hard to hear the details.
Yes I can imagine. For some of these tutorials you probably need to listen on your studio monitors or headphones.
As close as it doesn't make a difference, on most of the bits I would use
👍
How well do you think that Spectralayers version 10 does vs version 9?
Do you feel that the upgrade was worth it... can you tell a notable difference between the two versions for stem separation?
Well I only used SpectraLayers 8 so far. As I understand it, 8 and 9 did not differ very much regarding stem separation. Both from my own experience and from what I've seen from other people, version 10 is quite a big change in how well this works. It's never perfect of course, and never will be, but isolating Vocals, Bass and Drums is really amazing depending on what you want to use the stems for of course. Like I mentioned in the video.
Can you take a stereo drum mix and separate the individual drums?
Yes you can split up drums too. Kick, Snare and the rest I believe it was.
Hi am just starting using spectra layers when I open only shows by default vocal it won’t show other instruments how I can change that anyone can explain that appreciate tks
I don’t quite understand what you mean because when you open a stereo mix in SpectraLayers you should just hear the full mix. Maybe start with a video on the basics of SpectraLayers. I have one from a while back: ua-cam.com/video/nUHRBcZ5xhE/v-deo.htmlsi=ADOywn997GXseU22
CAn you record and mix Ur own vocals for any genre in spectral???
I don’t think I understand your question. What exactly do you want to do?
How come most all your audio on this end has an underlying distortion to it,I wonder if you sent up your audio just a couple db less if it would go away.
Not sure. Some tracks have more distortion then others but I don’t hear a general issue. More on the voice or also on the music?
My comments can be found at 13:26
So we agree :).
Spectral layers sucks
Why?