Stem separation: Remove vox to EQ guitar

[hieroglyph]

As is typically the case, I may be overcomplicating a very simple operation. I have a three track song. It is a live recording of vocals and guitar. There is a little guitar bleed in the vocal mic but I can live with it. However, there is way too much vocal bleed in the guitar mic. The third track is the same guitar performance but recorded through the pickup, therefore isolated.

If I want to keep the guitar mic track, can I use stem separation to remove the vocals from that track and just leave them out? Will there be leftover remnants of the removed vocal or will the other vocal track cover up said remnants?

I've considered just removing the guitar mic track altogether but it sounds more full to have it played with the DI guitar track.

Maybe this is as much a philosophical question as it is a technical question. Will stem separation allow me to EQ a guitar track minus the vocals? Or should I throw that track out completely and stick with the guitar DI and other vocal track? No current plans to overdub anymore instruments or vocals.

I have a mountain of recordings like this and re-recording them with vocals and guitar completely separated isn't an option. Thanks!

 

[lokeyfly]

Yes, that would be fine. You can re eq after you'd auto-stem separate. The kicker is knowone here could ever tell you what your results will be from auto-stem separate, because the process yields different results based on those recordings, and from each other. So you'll have to roll up your sleeves and just try it. Start separating those stems. Then see what you're left with. A little leakage shouldn't matter to re EQ, unless that leakage has harsh results such as plucky or noisy remnants. Even if so, you might be able to notch out the bad also with EQ.

In the time you wrote this question, you'd know how your mic and guitar would have separated. Not being funny, just honest my friend. Hey, let us know how it worked. Get 'er done.

 

[hieroglyph]

Yes, that would be fine. You can re eq after you'd auto-stem separate. The kicker is knowone here could ever tell you what your results will be from auto-stem separate, because the process yields different results based on those recordings, and from each other. So you'll have to roll up your sleeves and just try it. Start separating those stems. Then see what you're left with. A little leakage shouldn't matter to re EQ, unless that leakage has harsh results such as plucky or noisy remnants. Even if so, you might be able to notch out the bad also with EQ.

In the time you wrote this question, you'd know how your mic and guitar would have separated. Not being funny, just honest my friend. Hey, let us know how it worked. Get 'er done.

Thank you for your reply and I really appreciate the pep talk. I saw your post this morning and it energized me to go ahead and implement the stem separator. I am pleased with the results so far. As you said, the separated guitar needed a few more EQ adjustments but when played along with the vocal track and other guitar track, I think it sounds good.

I even used it in the same fashion to separate and EQ a clawhammer banjo track that had significant vocal bleed (different song). Now I need to rest my ears and check tomorrow to see if I am still happy with the results. Thanks again!

 

[hieroglyph]

Yes, that would be fine. You can re eq after you'd auto-stem separate. The kicker is knowone here could ever tell you what your results will be from auto-stem separate, because the process yields different results based on those recordings, and from each other. So you'll have to roll up your sleeves and just try it. Start separating those stems. Then see what you're left with. A little leakage shouldn't matter to re EQ, unless that leakage has harsh results such as plucky or noisy remnants. Even if so, you might be able to notch out the bad also with EQ.

In the time you wrote this question, you'd know how your mic and guitar would have separated. Not being funny, just honest my friend. Hey, let us know how it worked. Get 'er done.

Quick update. I used the stem separator more today. You were also correct that the process yields different results for different recordings. On some of the songs, the vocal comes out cleanly. On others, there is a lot more interference. I think it's because the singer's deepish voice occasionally clashes with some lower frequencies shared with the acoustic guitar and the vocal is somehow being pulled onto the guitar track at those frequencies (for lack of better vocabulary). EQ helps in some cases but not others. Since they were combined on a single track, I don't get to pick and choose which vocal remnants carry over and which stay put.

Another thing to consider is that the tool seems to be made for separating vocals, bass, and drums. The 'other' category is probably calibrated a little more broadly and the acoustic guitars and banjos are throwing it a curve ball, haha.

I want to add that I am not complaining, merely reporting my experience and analysis. Overall big picture, I am very happy with this tool in my toolbox and I think Presonus/S1 hit a home run by including it in the software.

 

[lokeyfly]

I want to add that I am not complaining, merely reporting my experience and analysis. Overall big picture, I am very happy with this tool in my toolbox and I think Presonus/S1 hit a home run by including it in the software.

Yeah, me too. Glad the stem separator is now there. I'm sure it will improve eventually, so let's use it for what we can get out if it. In a recent song of mine named "Ria Amazonez" (in my sig), I needed the stem separator to remove really noisy sounds over a toucan (bird). The noise were horrible camera shutter clicks and people talking over the bird chirping which I wanted to sound clean. The stem separator initially didn't work at all. I then had to hype the EQ at several crossiver points using the channel splitter and EQ. Then separate, and re adjust EQ. The results were good, but it took this extra processing. So I do think Studio One will add more flexibility to that, at some point.

Thank you hieroglyph for relaying back your findings! That really helps others in the same boat, and to the general questions of what the auto stem separator can or can't do. It can be coaxed further, but that's up to the users needs. Weather they'll try it is another story.
Glad it's somewhat working out for you.

 

[hieroglyph]

The stem separator initially didn't work at all. I then had to hype the EQ at several crossiver points using the channel splitter and EQ. Then separate, and re adjust EQ. The results were good, but it took this extra processing. So I do think Studio One will add more flexibility to that, at some point.

I checked out your Amazon video. I don't know if you've seen the film Koyaanisqatsi but your video kind of reminded me of it, albeit a 5 minute version. Your video is very cool and I completely agree with the premise of it being preservation of the rain forest, and Mother Nature in general.

Thank you again for your hot tips. I looked in the Studio One Toolbox page and found some videos on 'channels' and some on 'splitters' but I don't know if any of those look like tutorials on 'channel splitters'.

Do you mind please sharing a little more about the channel splitter? As I type this, I do not have S1 pulled up. Where is the channel splitter tool located? Is it part of ProEQ3 or is it something used in conjunction with the EQ? More specifically, how did you use the channel splitter? If my suspicions are correct, based on your descriptions, it sounds like this function may be exactly what I need to push these mixes into the end zone.

 

[lokeyfly]

Absolutely.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

0:00 Marcus shows the process of channel splitting with Pro Tools and basic operation as an intro to what you'll be doing.

1:10 Marcus's introduces Studio One's Splitter feature, along with its three ways to split the channel.

3:31 Marcus sets the Splitter for Crossover.
This is what I was referring to. Think of the crossover as no different than if you were running an active crossover to your home, Studio, or theater audio system. The crossover is acting like a steep frequency divider (varied balance) or cut-off (hard to one side balance) for desired frequencies to pass though.

If you need your track to be or stay in stereo similar to a crossover, you could instead use a HPF or LPF (high/low pass filter). Your choice.

Another EQ crossover you might want to explore is Studio One's Multiband compressor. This also provides frequency crossover points at 4 locations! That can be pretty effective isolating any issues you're trying to remove, or pass through.
This all in an effort to allow you to isolate some sounds ir stemming for your project.
Hope this helps.

Marcus has good tips as do some of our excellent folks here in the forum.
Hope this helps..

 

[hieroglyph]

Yeah, me too. Glad the stem separator is now there. I'm sure it will improve eventually, so let's use it for what we can get out if it. In a recent song of mine named "Ria Amazonez" (in my sig), I needed the stem separator to remove really noisy sounds over a toucan (bird). The noise were horrible camera shutter clicks and people talking over the bird chirping which I wanted to sound clean. The stem separator initially didn't work at all. I then had to hype the EQ at several crossiver points using the channel splitter and EQ. Then separate, and re adjust EQ. The results were good, but it took this extra processing. So I do think Studio One will add more flexibility to that, at some point.

Thank you hieroglyph for relaying back your findings! That really helps others in the same boat, and to the general questions of what the auto stem separator can or can't do. It can be coaxed further, but that's up to the users needs. Weather they'll try it is another story.
Glad it's somewhat working out for you.

Thank you for taking the time to help me. Unfortunately, I took my newfound knowledge and may have made the issue worse.

The above bolded part seemed like a simple enough task. I took one song and using the video you posted, I was able to figure out how to implement the channel splitter. The track I split had vocal and guitar. I added EQ to one of the split branches and attempted to remove or decrease the vocal frequencies so maybe the stem separator would work more cleanly. That didn't work. (What do you mean by 'hype the EQ'?)

3:31 Marcus sets the Splitter for Crossover.
This is what I was referring to. Think of the crossover as no different than if you were running an active crossover to your home, Studio, or theater audio system. The crossover is acting like a steep frequency divider (varied balance) or cut-off (hard to one side balance) for desired frequencies to pass though.

If you need your track to be or stay in stereo similar to a crossover, you could instead use a HPF or LPF (high/low pass filter). Your choice.

Another EQ crossover you might want to explore is Studio One's Multiband compressor. This also provides frequency crossover points at 4 locations! That can be pretty effective isolating any issues you're trying to remove, or pass through.
This all in an effort to allow you to isolate some sounds ir stemming for your project.
Hope this helps.

Marcus has good tips as do some of our excellent folks here in the forum.
Hope this helps..

Now I have two problems. I think my first issue is that I don't quite understand what you are talking about when you describe 'crossover'. I see the part about how "crossover is acting like a steep frequency divider (varied balance) or cut-off (hard to one side balance) for desired frequencies to pass though." I'm still having trouble wrapping my head around the concept of crossover and what to do about it. Your description of crossover in the context of home or studio audio systems does not resonate with me as I don't know much about wiring audio systems.

To add insult to injury, the second problem is that somewhere somehow I did something and now I can no longer get the stem separator to work with that song. At all. After attempting to reverse engineer what I'd done with the channel splitter, I even went backwards with my song and imported the original raw tracks into a new project just so I could see if I could get the stem separator to work again. Nada. I right click the track, choose 'separate stems', choose 'vocal' and 'other', the window pops up for a minute or so as the stem separator engine revvs up, and then nothing. The track is still highlighted but it is not broken into stems.

It got to the point where I didn't want to spend all day banging my head so I hit pause on that project and went to work on other songs that don't need stem separation.

I am not giving up and will try again today, I just wish there were some tutorial videos that showed how to use the channel splitter in tandem with EQ the way you used it. Most of the youtube reviews of S1's stem separator tool conclude that it doesn't work very well, but none of them are using the channel splitter or EQ to finesse the track before plugging in the stem separator. You seem to have cracked the code before anyone else.

Sorry for the long vent rant. I'm sure I am overcomplicating one or more steps in this process. Thank you for your time.

 

[lokeyfly]

Hyping the frequency is exaggerating the equalization to allow each band to have greater differences from one another (as apposed to simply boosting or dropping bass, mid, and treble). So for instance as in your case, the auto stemming was not separating your stems to desired separation of separate tracks. "Hyping" frequencies would be changing EQ to some alternate bias. Like making treble EQ higher, bass frequency lower, or ducking (dropping out some of the EQ at some range as in low to mid frequencies) or even notching (gapping) out frequences. Then re introducing your changes into the stem separator for different results. Possibly better, possibly not.

Thank you for your time.

You're welcome, but I'm afraid I'm probably not breaking this down for you in the way you need. I could suggest a really great tutor if you want to go that route. Let me know. He's a member in this forum and is very approachable. His name is Johnny Geib. Here is his link at Home Studio Trainer. He'll at least be able to explain the steps with you.

Home studio help and support | HST-HomeStudioTrainer.com

At www.hst-homestudiotrainer.com Learn how to build and operate your own home recording studio. Also learn the ins and outs of the easiest DAW on the market, Presonus Studio One Artist or Professional

www.hst-homestudiotrainer.com

I haven't cracked any codes yet. Just flying by the seat of my pants, but I think your getting the process. You just have to try different things which is why I offered a few options (Splitter, multiband compressor, etc.). If the same EQ range resides in something like a guitar and vocals and can't be separated. Then it may not be possible to provide further separation. So really allow your ears and a little help from a frequency analyzer, to understand what it is, you're trying to separate. Studio One has a fairly useful frequency analyzer. It's funny how we musicians/producers have become technical gurus, to get the job done.

No worries, you're not over complicating anything. Auto stemming is a new S1 feature, and it's both a little limited in its control compared to some other auto separators, and the process from our position is still not fully explored when using it. So experiment, and also light a few candles. : )

 

[hieroglyph]

Hyping the frequency is exaggerating the equalization to allow each band to have greater differences from one another (as apposed to simply boosting or dropping bass, mid, and treble). So for instance as in your case, the auto stemming was not separating your stems to desired separation of separate tracks. "Hyping" frequencies would be changing EQ to some alternate bias. Like making treble EQ higher, bass frequency lower, or ducking (dropping out some of the EQ at some range as in low to mid frequencies) or even notching (gapping) out frequences. Then re introducing your changes into the stem separator for different results. Possibly better, possibly not.


You're welcome, but I'm afraid I'm probably not breaking this down for you in the way you need. I could suggest a really great tutor if you want to go that route. Let me know. He's a member in this forum and is very approachable. His name is Johnny Geib. Here is his link at Home Studio Trainer. He'll at least be able to explain the steps with you.

Home studio help and support | HST-HomeStudioTrainer.com

At www.hst-homestudiotrainer.com Learn how to build and operate your own home recording studio. Also learn the ins and outs of the easiest DAW on the market, Presonus Studio One Artist or Professional

www.hst-homestudiotrainer.com

I haven't cracked any codes yet. Just flying by the seat of my pants, but I think your getting the process. You just have to try different things which is why I offered a few options (Splitter, multiband compressor, etc.). If the same EQ range resides in something like a guitar and vocals and can't be separated. Then it may not be possible to provide further separation. So really allow your ears and a little help from a frequency analyzer, to understand what it is, you're trying to separate. Studio One has a fairly useful frequency analyzer. It's funny how we musicians/producers have become technical gurus, to get the job done.

No worries, you're not over complicating anything. Auto stemming is a new S1 feature, and it's both a little limited in its control compared to some other auto separators, and the process from our position is still not fully explored when using it. So experiment, and also light a few candles. : )

Once again, thank you for your time. I dug back in today and I had better luck than yesterday. The first thing I did was import the original WAVs into a new project and check the stem separator. Works. Then I saved and closed and opened another new project but I tested the channel splitter this time. I found that if I mixdown to stereo, the channel splitter works better on a stereo track (facepalm). When I attached the splitter, I created 3 channels instead of 2, which is something else I did differently than yesterday. I was able to adjust the frequencies on all three parts so thought I was almost there. When I went to apply the stem separator to that track, nothing happened. Same as yesterday. But I know stem separator works so instead of messing around with the channel splitter, I adjusted some frequencies in the ProEQ, pre-stem-separation. Then I separated the stems and again applied some EQ to the track. Seems to have worked a little, I will check again tomorrow.

TL/DR: Could not get the stem separator to work after applying the channel splitter. Used regular EQ instead of splitter and is seeing better results than previously.

Thanks again for your help, lokeyfly. And thank you for the link to the tutor. If I get in too deep, I might give him a shout.