BlueberryAlien
VDJ stems arent the worst but they sound weird a lot of the time and having to circumvent the stem weirdness is getting tiresome, so I have to use an external application to create better stems and import with the Stem Creator. You should just collab with a company thats good at this
Posted Wed 30 Sep 26 @ 4:39 pm
DJ VinylTouch
This is a comment that shows a lack of understanding of tradeoffs.
Some info for you to digest in understanding the state of affairs in DJ stem technology
Where "Good enough" is probably defined as usable enough for no obvious complaints quality issues in the average case.
Once you understand that, then you'd understand that tradeoffs have to be made in order to achieve that goal.
Stem tools that are not used for realtime, light and fast separation as a focus, can focus on making the best separation possible, so it doesn't make sense to compare them to DJ software as they have different goals/restrictions.
With respect to DJ software - ALL of them made compromises to get the product they have (you could just look at the tons of stem comparison videos for evidence). Some may have focused more on clear drums, some vocals, some otherwise, so you'll always end up with cases where one song may be best separated in a certain aspect vs others.
It doesn't matter if you licenced your algorithm from someone or not, that is the fact that you can't get around (and licencing from someone else is worse, for obvious reasons - they control a part of the product you are delivering and you are at the whim of that company).
I, and many others, are pretty satisfied with the Stems 2.0 performance, especially as we know the limits and what one would use it for.
If you are trying to get perfect separation for all of your tracks, then this is not the tool to do it (and furthermore, no DJ software is the right tool for that job because they all compromise) - stem creation with producer stems or other of better quality is where it's at. Anything otherwise would require more resources (better spec'ed machine, much larger model, more time), which gives probably neglible return compared to the primary goal.
Some info for you to digest in understanding the state of affairs in DJ stem technology
Quote :
The aim of stem usage in DJ software is not to be perfect, it's to be performant (especially in terms of time to separate and resources used) and good enough for most use cases
The aim of stem usage in DJ software is not to be perfect, it's to be performant (especially in terms of time to separate and resources used) and good enough for most use cases
Where "Good enough" is probably defined as usable enough for no obvious complaints quality issues in the average case.
Once you understand that, then you'd understand that tradeoffs have to be made in order to achieve that goal.
Stem tools that are not used for realtime, light and fast separation as a focus, can focus on making the best separation possible, so it doesn't make sense to compare them to DJ software as they have different goals/restrictions.
With respect to DJ software - ALL of them made compromises to get the product they have (you could just look at the tons of stem comparison videos for evidence). Some may have focused more on clear drums, some vocals, some otherwise, so you'll always end up with cases where one song may be best separated in a certain aspect vs others.
It doesn't matter if you licenced your algorithm from someone or not, that is the fact that you can't get around (and licencing from someone else is worse, for obvious reasons - they control a part of the product you are delivering and you are at the whim of that company).
I, and many others, are pretty satisfied with the Stems 2.0 performance, especially as we know the limits and what one would use it for.
If you are trying to get perfect separation for all of your tracks, then this is not the tool to do it (and furthermore, no DJ software is the right tool for that job because they all compromise) - stem creation with producer stems or other of better quality is where it's at. Anything otherwise would require more resources (better spec'ed machine, much larger model, more time), which gives probably neglible return compared to the primary goal.
Posted Wed 30 Sep 26 @ 4:53 pm
kradcliffe
I'm perfectly happy with VDJ stems 2.0 for DJ use and being able to load them live at 27x at a gig or 42x at home.
All you ever seem to do is have a moan about something.
All you ever seem to do is have a moan about something.
Posted Wed 30 Sep 26 @ 5:04 pm
BlueberryAlien
DJ VinylTouch wrote :
This is a comment that shows a lack of understanding of tradeoffs.
Some info for you to digest in understanding the state of affairs in DJ stem technology
Where "Good enough" is probably defined as usable enough for no obvious complaints quality issues in the average case.
Once you understand that, then you'd understand that tradeoffs have to be made in order to achieve that goal.
Stem tools that are not used for realtime, light and fast separation as a focus, can focus on making the best separation possible, so it doesn't make sense to compare them to DJ software as they have different goals/restrictions.
With respect to DJ software - ALL of them made compromises to get the product they have (you could just look at the tons of stem comparison videos for evidence). Some may have focused more on clear drums, some vocals, some otherwise, so you'll always end up with cases where one song may be best separated in a certain aspect vs others.
It doesn't matter if you licenced your algorithm from someone or not, that is the fact that you can't get around (and licencing from someone else is worse, for obvious reasons - they control a part of the product you are delivering and you are at the whim of that company).
I, and many others, are pretty satisfied with the Stems 2.0 performance, especially as we know the limits and what one would use it for.
If you are trying to get perfect separation for all of your tracks, then this is not the tool to do it (and furthermore, no DJ software is the right tool for that job because they all compromise) - stem creation with producer stems or other of better quality is where it's at. Anything otherwise would require more resources (better spec'ed machine, much larger model, more time), which gives probably neglible return compared to the primary goal.
Some info for you to digest in understanding the state of affairs in DJ stem technology
Quote :
The aim of stem usage in DJ software is not to be perfect, it's to be performant (especially in terms of time to separate and resources used) and good enough for most use cases
The aim of stem usage in DJ software is not to be perfect, it's to be performant (especially in terms of time to separate and resources used) and good enough for most use cases
Where "Good enough" is probably defined as usable enough for no obvious complaints quality issues in the average case.
Once you understand that, then you'd understand that tradeoffs have to be made in order to achieve that goal.
Stem tools that are not used for realtime, light and fast separation as a focus, can focus on making the best separation possible, so it doesn't make sense to compare them to DJ software as they have different goals/restrictions.
With respect to DJ software - ALL of them made compromises to get the product they have (you could just look at the tons of stem comparison videos for evidence). Some may have focused more on clear drums, some vocals, some otherwise, so you'll always end up with cases where one song may be best separated in a certain aspect vs others.
It doesn't matter if you licenced your algorithm from someone or not, that is the fact that you can't get around (and licencing from someone else is worse, for obvious reasons - they control a part of the product you are delivering and you are at the whim of that company).
I, and many others, are pretty satisfied with the Stems 2.0 performance, especially as we know the limits and what one would use it for.
If you are trying to get perfect separation for all of your tracks, then this is not the tool to do it (and furthermore, no DJ software is the right tool for that job because they all compromise) - stem creation with producer stems or other of better quality is where it's at. Anything otherwise would require more resources (better spec'ed machine, much larger model, more time), which gives probably neglible return compared to the primary goal.
Well guess why Im making this post... It's often not good enough. Whereas with frequencies you can predict how it's going to sound like, with stems it's always a matter of taking a guess, and I'd say a lot of the time you think it's gonna sound like something but it doesnt at all and it is artifacted/bleeding to all hell. Well thanks to you guys, the devs will look into making shoutout v2 instead of perfecting stems or collabing with some company dedicated to creating good stems.
Posted Wed 30 Sep 26 @ 10:03 pm
BlueberryAlien
kradcliffe wrote :
I'm perfectly happy with VDJ stems 2.0 for DJ use and being able to load them live at 27x at a gig or 42x at home.
All you ever seem to do is have a moan about something.
All you ever seem to do is have a moan about something.
Get your ears checked then. I have to import stems from Fadr due to VDJ stems not being good enough. Theyre fine for live mixing but when you want an impeccable recorded mix, it's not good enough.
Posted Wed 30 Sep 26 @ 10:04 pm
DJ VinylTouch
You're answer is "just collab with a company".
You don't see why that doesn't make sense? You're just making a blanket statement (which doesn't really point out or solve the problem), instead of pointing out what problems you have so they can fix it.
As for it being guess work, that's pretty much your opinion - I've never guessed with using it. i know what set of frequencies should appear/disappear based on stem type activation, and I know expectation of limits. Pretty much use it every week without issues/noticeable bleed in the cases (and I'm not the only one either).
It's not that it can't improve, it's just that it sounds like you're ranting without an actual problem to solve, and saying that the problem is solved by buying some new product off the shelf, which isn't useful.
You don't see why that doesn't make sense? You're just making a blanket statement (which doesn't really point out or solve the problem), instead of pointing out what problems you have so they can fix it.
As for it being guess work, that's pretty much your opinion - I've never guessed with using it. i know what set of frequencies should appear/disappear based on stem type activation, and I know expectation of limits. Pretty much use it every week without issues/noticeable bleed in the cases (and I'm not the only one either).
It's not that it can't improve, it's just that it sounds like you're ranting without an actual problem to solve, and saying that the problem is solved by buying some new product off the shelf, which isn't useful.
Posted Wed 30 Sep 26 @ 10:08 pm
BlueberryAlien
Well clearly then we dont play the same kind of music. If they think it cant be done all they have to say is No it cant be done and thats that. But they used to say they couldnt EVER get spotify on board so you never know.
Posted Thu 01 Oct 26 @ 3:15 pm
kradcliffe
Are you playing music in live venues as a DJ or just at home?
Posted Thu 01 Oct 26 @ 3:18 pm
DJ VinylTouch
It can be done. The actual problem is, whether or not anything needs to be done at all (i.e. it is good enough), and if it is improved, at what cost does the improvement come?
As said before, if you are looking for perfection of separation, then you're out of luck. There is no perfect separation tool, and DJ software has to be light enough to run on average machines + fast enough for quick decision mixing (as described before).
The second point is obvious.
The first point (does anything need to be done at all) needs example songs, otherwise you're just saying things (e.g. you play different songs, etc).
Like your music isn't so special that nobody plays it - so if there really is an issue you should readily be able to provide examples, otherwise it's just a rant with no real point.
As said before, if you are looking for perfection of separation, then you're out of luck. There is no perfect separation tool, and DJ software has to be light enough to run on average machines + fast enough for quick decision mixing (as described before).
The second point is obvious.
The first point (does anything need to be done at all) needs example songs, otherwise you're just saying things (e.g. you play different songs, etc).
Like your music isn't so special that nobody plays it - so if there really is an issue you should readily be able to provide examples, otherwise it's just a rant with no real point.
Posted Thu 01 Oct 26 @ 4:25 pm