• Hi and welcome to the Studio One & Studio Pro User Forum!

    Please note that this is an independent, user-driven forum and is not endorsed by, affiliated with, or maintained by Fender. Learn more in the Welcome thread!

Stem Separation?

Ken Morgan

Member
Please educate me on these things, Looks like a really popular 'feature' in DAWs and as standalone applications, in the real world, what use does it have, other than an attempt to undo a final mix some else painstakingly did for whatever reason and reassemble it differently than originally intended.

Put another way, don't these stem separation tools offer some fairly easy ways to sidestep copyrighted materials? Please note I am NOT trying to start an argument, Im trying to understand the actual real world uses for stem separation apps and the ethical responsibilities users have in deploying them

Thanks
 
Obviously stem separation can have unscrupulous uses, but so can my kitchen knife. I use it (stem separation, not the knife!) for analysis and for breaking down old recordings. I'm a pretty rubbish guitar player, and it's quite helpful to isolate guitar parts so that I can make an attempt at learning them. Actually, scratch 'guitar' - it applies to keyboards and vocals too! Part of my music time is spent creating choral arrangements mockups for local choirs, and getting into other covers' harmonies can be really useful for that. Stem separation can help, and if I can get a clean enough stem I can analyse it further with Melodyne Studio.
 
It can be a very effective tool for cleaning up vocal/bass/... tracks that suffer from recording noise or bleed. That way you use stem separation as a smart filter rather than trying to micro-edit the track yourself. And it only costs seconds so if the result is below par then the painful road is still there to take.
 
Please educate me on these things, Looks like a really popular 'feature' in DAWs and as standalone applications, in the real world, what use does it have, other than an attempt to undo a final mix some else painstakingly did for whatever reason and reassemble it differently than originally intended.

Put another way, don't these stem separation tools offer some fairly easy ways to sidestep copyrighted materials? Please note I am NOT trying to start an argument, Im trying to understand the actual real world uses for stem separation apps and the ethical responsibilities users have in deploying them

Thanks
I had one of my songs recorded years ago on an Tascam 388. Only had 8 tracks and drum sync took up 1 track.
I accidentally bounced the Bass and Guitar onto one track and did not realize it.
With Stem separation I was finally able to Isolate the Bass on it own Track (y)🖖
 
Kind of a specialized use of separation, but . . .

In 1972 I recorded a song of mine on a TEAC quarter-track deck, bouncing left channel to right several times to add all the instruments and vocals Needless to say the end result was horrible, more noise than music. But I kept the recording for the next 50 years, and one day I uploaded the bad recording to Suno, prompting it to "cover" the song, enhancing the sound but no other changes. In about two minutes I had a lovely version of my song with a full orchestra. But the AI "singer" couldn't get the lyrics right, and here's where stem separation enters.

I broke the recording into stems, deleted the AI voice, and sang my song over the Suno-created backing track. Back in '72 I could not have produced such a slick track. Sadly, 50 years on, my voice is not up to the task, but at least I have a listenable recording of the song with all the correct words.
 
I have a recording my band did in the late 60s. The singer was sick, so his voice ran out of gas at the end of phrases and he had some pitch issues. The song was released anyway on the album Puzzle, but it always bothered me that his voice didn't sound anything like it did live. He died many years ago, so there was no way to ever cut a new version and make it right.

With stem separation, I extracted the vocal and made a copy. One extract was flipped out of phase to cancel the original vocal. That was key to making this process work (AKA Karaoke, LOL). Then, I processed the other vocal with a little bit of gain envelope action and some judicious pitch correction. I mixed that with the stereo track where the vocal had been removed, and the song ended up sounding like what it sounded like live, with him at full strength.

Stem separation basically revived him long enough to do the vocal right, over half-a-century later.
 
You win this Battle of Reincarnated Recordings. Mine was only 54 years old

LOL!

This recording was done at one of those Big Studios, so I tried hunting down the multitracks. Of course, they're long gone. At least you kept yours! I had to work with the stereo version that hade been remastered years later for a CD re-release, so it would not have been possible to fix it without stem extraction.
 
Thanks for some real world application examples. FWIW in my own works when something was just done wrong, I tend to just re-do it top to bottom as the tools available to me now are (usually) so much more advanced than what was available when originally done.
 
I tend to just re-do it top to bottom
Ken - Some of us will geek around endlessly with a bad track trying to salvage parts and replace others. I'm with you -- I feel like I can always create something better from scratch. In my example above I felt like I had no choice but to attempt a salvage operation.🙃
 
What's also interesting is that the sum of the separated stems equals the original. So blending the stems allows you to soften conspicuous artefacts.
 
What's also interesting is that the sum of the separated stems equals the original. So blending the stems allows you to soften conspicuous artefacts.

Clever!

In my example above I felt like I had no choice but to attempt a salvage operation

Same with me - the singer had died. But also, let's not forget about using audio-to-MIDI in conjunction with stem separation. I have some old music with drum sounds that are nowhere near as good as what's available today. Extracting the drum audio, then using audio-to-MIDI to create a MIDI track, let me trigger much better drum sounds while retaining the original performance.

I think this thread indicates that stem extraction isn't just a one-trick pony(y) There are probably other applications we'll all discover in the years ahead.
 
in the real world, what use does it have, other than an attempt to undo a final mix some else painstakingly did for whatever reason and reassemble it differently than originally intended.
Looks like you got your answer. : )
There are numerous reasons and more. Now, if I could only get my hands on a Studer A820, Revox A77 MkIII, or B77. I could remix, or bring back those 1980 recordings of our band. Recorded in our studio, and Plaza Studio (old NBC Studio) over Radio City Music Hall.
 
In the times of remixes and mashups stem separation is a godsend (but not for everyone). Recently, it was even propagated by Ski Oakenfull, a teacher at the British Pointblank Music School, in a YouTube video as a suitable source for vocal stems for remixes.
 
bring back those 1980 recordings of our band
A lot of us harbor those wistful thoughts, but most of those old tapes are unplayable, because the rubberized backing has turned to sticky goo.
 
TBH I think the use cases described in this thread are very much edge cases rather than the mainstream use for stems, which is remixing and sampling. At the "top level" people are using parts of other songs in hit records, presumably with proper sample clearance, but lower down the tree it is very common for folks to borrow elements of other songs without any clearance. I guess that only becomes a problem if they suddenly have a viral hit on their hands! This stuff has been going on for decades but stem separation makes it a whole lot easier and means you can get access to a drum or instrument loop that just was not possible previously.

The more creative artists use the sample as an instrument, chopping it into segments and rearranging/repitching it but there are plenty of people just using elements of other songs as is, for a hook or for a counterpoint. That's why stem separation inside MPC hardware is such a big deal. You can now do this stuff directly on the hardware.

Of course, most of the folks doing stuff like this are probably using a different DAW...
 
I have no idea what Ableton’s EDM and Hip Hop creators are using it for, but sampling has been going on, legally or otherwise, for many, many years! Stem separation may make it easier to get at the sounds they want, but in a way brings it closer to sample-based synthesis.

Another use that is probably more mainstream, and is already wrecking the old music we love, is the ‘remaster’ trend, where the characterful but imperfect vocals of the original recordings are separated out, pitch corrected within an inch of their lives, and recompiled within a clinically sterile but competitively loud new mix.
 
Isn't it in the nature of tools that in the wrong hands every tool can be dangerous? Then let's not blame the tool but blame the hands. Everyone is (or should be) personally responsible for his or her own actions, so that when the hammer comes down (;)) there's nowhere to hide.
 
Correct, and since there are many OpenSource solutions out there (providing even better quality than the original tool in v7) it’s safe to say: “Pandora’s box is open”.
 
Isn't it in the nature of tools that in the wrong hands every tool can be dangerous? Then let's not blame the tool but blame the hands. Everyone is (or should be) personally responsible for his or her own actions, so that when the hammer comes down (;)) there's nowhere to hide.

I'd suggest this is a wide, but accurate assessment. The hammer used to build homes, offices, hospitals, etc can and has been used to commit acts of violence, and no one ever blames the hammer.

But no one glorifies the hammer, either, creating and fine tuning ways that more times than not, dance around the legalities of its existence and usage. This notion was the gist of my initial thought. No one buys a vehicle with the purpose of getting drunk and killing someone in a DUI incident; (almost) no one buys a firearm with the sole purpose of killing someone else; etc.

The very cool examples provided above are likely outliers. I'd hazard a guess the folks who use stem separation to rip a phrase, a sound, whatever, aren't going to post their methods or results here

BTW - maybe its just me, but I'm thinking this is a good conversation
 
Back
Top