Hi, I'm Rekordbox ans Serato user to years and very interested in Virtual DJ,
Prepared Stems with Virtual DJ sounds very good. But I won't prepare all stems.
On Windows 11 Lenovo T16 / i17 32GB ram, I can play and prepare stems with lower quality witout hearing anything.
But with my FLX-10 connected I can hear glitches - and the music can even stop.
PLEASE - its not my computer - its not my usb cable.
No - itrs Virtual DJ - Because I DONT HAVE THESE PROBLEMS WITH SERATO OR REKORDBOX.
I tryed all possibiites on configuration : with/witout GPU, lower priority, lower quality - I tried tu tune up my PC - USB power , etc - but nothing helps.
THe best is without GPU, Lower priority, lowaer quality, but there are always hearable gliteches when loading a track. Its just like all the CPU concentrates to compute next track stems - but I would like to compute with even lesser priority .
QUESTION :
1. - why should have these problems only connecting a controlloer ? This is not usable in live, because we cannot choose either or not computing next track stems - so there are ALWAYS glitches when loading a track
2 . why Serato or Rekordbox don't have this problem and you do ?
This is a severe bug report, plase tell me if there is already a ticket mentioning this issue.
Very best regards
Please reaise
Prepared Stems with Virtual DJ sounds very good. But I won't prepare all stems.
On Windows 11 Lenovo T16 / i17 32GB ram, I can play and prepare stems with lower quality witout hearing anything.
But with my FLX-10 connected I can hear glitches - and the music can even stop.
PLEASE - its not my computer - its not my usb cable.
No - itrs Virtual DJ - Because I DONT HAVE THESE PROBLEMS WITH SERATO OR REKORDBOX.
I tryed all possibiites on configuration : with/witout GPU, lower priority, lower quality - I tried tu tune up my PC - USB power , etc - but nothing helps.
THe best is without GPU, Lower priority, lowaer quality, but there are always hearable gliteches when loading a track. Its just like all the CPU concentrates to compute next track stems - but I would like to compute with even lesser priority .
QUESTION :
1. - why should have these problems only connecting a controlloer ? This is not usable in live, because we cannot choose either or not computing next track stems - so there are ALWAYS glitches when loading a track
2 . why Serato or Rekordbox don't have this problem and you do ?
This is a severe bug report, plase tell me if there is already a ticket mentioning this issue.
Very best regards
Please reaise
Posted Sun 10 May 26 @ 4:28 pm
does your system meets the specs for Stems 2.0 i do not see a video card in your post
https://virtualdj.com/wiki/Minimum%20system%20requirements.html
see the additional for stems 2.0
what is your power profile set to?
has this machine been optimized for DJing?
https://virtualdj.com/wiki/Minimum%20system%20requirements.html
see the additional for stems 2.0
what is your power profile set to?
has this machine been optimized for DJing?
Posted Sun 10 May 26 @ 5:14 pm
No - only intel chipset - but I'm quite surprised that real time Stems even lower quality need a gamer PC -
Serato real time stems are better qualiy than VDJ real time stem lower quality - and serato runs fine with my computer.
My conclusion : I can't use VDJ because of my computer - thats a shame.
Serato real time stems are better qualiy than VDJ real time stem lower quality - and serato runs fine with my computer.
My conclusion : I can't use VDJ because of my computer - thats a shame.
Posted Mon 11 May 26 @ 10:39 pm
Serato introduced stems well after VirtualDJ and DJay Pro's release, and AFAIK there was no "lower quality" setting using lower specs on their side, so the only real comparison is VirtualDJ Stems 2.0 (which came out as an improvement).
As already described before, the specs requirements for everything are detailed here:
https://virtualdj.com/wiki/Minimum-system-requirements.html
The PC optimization guide is described here:
https://virtualdj.com/wiki/PC%20Optimize.html
What is the specs of your computer (graphics card included)?
I believe there is a way to force using your CPU for the job (devs please correct me if I'm wrong here).
Did you try a fresh settings.xml to see id the problem persiss (I believe the correct stems level is suggested based on your computer specs)?
As already described before, the specs requirements for everything are detailed here:
https://virtualdj.com/wiki/Minimum-system-requirements.html
The PC optimization guide is described here:
https://virtualdj.com/wiki/PC%20Optimize.html
What is the specs of your computer (graphics card included)?
I believe there is a way to force using your CPU for the job (devs please correct me if I'm wrong here).
Did you try a fresh settings.xml to see id the problem persiss (I believe the correct stems level is suggested based on your computer specs)?
Posted Mon 11 May 26 @ 10:53 pm
Stephen kay wrote :
No - only intel chipset - but I'm quite surprised that real time Stems even lower quality need a gamer PC -
Serato real time stems are better qualiy than VDJ real time stem lower quality - and serato runs fine with my computer.
My conclusion : I can't use VDJ because of my computer - thats a shame.
Serato real time stems are better qualiy than VDJ real time stem lower quality - and serato runs fine with my computer.
My conclusion : I can't use VDJ because of my computer - thats a shame.
Real time stems "lower quality" (or Stems 1.0) do not require a gaming rig to run properly.
However the quality of the stems separation is LOWER and audio artifacts are to be expected.
"Higher quality" stems separation (or Stems 2.0) still doesn't require a gaming rig to run properly, if you are willing to prepare the stems of your files in advance.
If you want to use both "Higher quality" and "Real Time" stems separation, at the same time, then YES you need a powerful machine and that's simply a limitation of the technology used that you cannot bypass.
Neither of those things is a "bug" too..
Lower quality stems separation has some artifacts here and there. It left more to be desired because of this, and thus Stems 2.0 ("high quality") were introduced.
In theory Stems 2.0 can also be further improved. But the time needed for the separation increases exponentially. For 5% quality improvement the time would increase by 20% or more.
For 10% quality improvement the time would increase by 80% or more.. e.t.c.
Don't get what I'm about to say wrong, but no matter the software, if you want to race a F1 track, you need a F1 car.
Using stems (even with competitor software) is racing a F1 track.
I know (because I do various tests all the time) how much time Serato or Rekordbox needs for stems separation, and I also know the "quality" of their separation.
If you cherry pick tracks, you can claim that Serato does the same job as VirtualDJ in slightly less time on a less powerful system.
BUT: If you just throw a random folder with 100 tracks on them, VirtualDJ does a far better job. Even if it takes slightly longer, even if it needs a more powerful machine. And we are fine with that.
Posted Tue 12 May 26 @ 7:35 am
PhantomDeejay wrote :
In theory Stems 2.0 can also be further improved. But the time needed for the separation increases exponentially. For 5% quality improvement the time would increase by 20% or more.
For 10% quality improvement the time would increase by 80% or more.. e.t.c.
For 10% quality improvement the time would increase by 80% or more.. e.t.c.
I have been feeling, at times, that I would love, say a, "Stems 4.0" which would be say be 50% better, (and maybe take 4000% longer to compute) for those rare, (1 in 500 tracks) where the vocal can not be isolated well.
The issue is, either, the vocal is not clear, OR it contains some remnants of the melody.
These vary rare tracks could be precomputed, in advance, when time is not an issue.
(and when a clear acapllea is desired, or needed for a vocal sampler)
And of course would never be used live, as Stems 2.0 is amazing for live stems separation (the best).
I can add it to the "wishes" forum, if you feel that is doable, and something that has a chance of being added.
Posted Wed 13 May 26 @ 1:01 am
I think for that, you'd need a very high quality master that's never been subjected to (file format) compression, where the voice is still intact behind the other sounds - even then, the processing required to remove the other sounds is (IMO) always going to cause artifacts of some kind.
Maybe an AI system could create an artificial version with no artifacts, but then people could argue it's not genuine?
Maybe an AI system could create an artificial version with no artifacts, but then people could argue it's not genuine?
Posted Wed 13 May 26 @ 9:28 am
I think there is still a misunderstanding about my issue, so let me clarify it more precisely.
VirtualDJ currently has two different stem systems:
* Stems 1.0
* Stems 2.0
These are NOT equivalent.
Stems 2.0 are designed to be precomputed/analyzed in advance. They are not intended for realtime generation during performance.
Stems 1.0 can be generated in realtime.
In the VDJ settings, Stems 1.0 are explicitly labeled as “Lower Quality”. This is not my opinion — this is the terminology used by VirtualDJ itself.
Now, let’s compare this behavior with other DJ software:
* Rekordbox also has two stem quality modes.
* Serato also has both realtime stems and precomputed/prepared stems, even if this is less explicitly exposed in the UI.
Here is the important point:
On Rekordbox:
* “Lower quality” stems are not really good enough for professional stage use.
* “Higher quality” stems sound better, but they are extremely heavy because they are always computed from the beginning of the track and cannot realistically be prepared for a large library unless using a very powerful gaming-class machine.
On Serato:
* realtime stems are actually very usable on stage.
* They are not as good as VDJ Stems 2.0, but they are absolutely sufficient for live performance.
* Most importantly: they work WITHOUT audio glitches on my system.
Now my actual issue:
On VirtualDJ, Stems 2.0 sound excellent. However, they require very long preprocessing on my system because I do not own an RTX 4070+ / high-end eGPU setup. Therefore, I cannot realistically precompute my entire music library.
So naturally, I expected Stems 1.0 (“Lower Quality”) to behave similarly to Serato realtime stems:
* lower quality,
* but usable live,
* and generated in realtime without audio interruptions.
But this is NOT what happens.
On my system:
* when VirtualDJ generates Stems 1.0 in realtime for a track that has not yet been analyzed,
* loading that track produces audible glitches,
* and sometimes even audible audio cuts/dropouts.
This is the core issue.
And to be very clear:
this is NOT a general hardware instability problem.
Because:
* With Serato:
* realtime stems work correctly,
* no glitches,
* quality is acceptable for stage use.
* The ONLY time Serato glitches is when my laptop is unplugged and Windows switches to aggressive power-saving behavior affecting USB/CPU scheduling.
But when plugged in and in maximum performance mode:
* Serato realtime stems = stable.
* VirtualDJ realtime Stems 1.0 = audible glitches/dropouts.
So from my perspective:
* this is not simply “expected CPU load”,
* this appears to be a specific realtime stem processing or audio scheduling issue inside VirtualDJ.
That is the issue I am trying to report.
VirtualDJ currently has two different stem systems:
* Stems 1.0
* Stems 2.0
These are NOT equivalent.
Stems 2.0 are designed to be precomputed/analyzed in advance. They are not intended for realtime generation during performance.
Stems 1.0 can be generated in realtime.
In the VDJ settings, Stems 1.0 are explicitly labeled as “Lower Quality”. This is not my opinion — this is the terminology used by VirtualDJ itself.
Now, let’s compare this behavior with other DJ software:
* Rekordbox also has two stem quality modes.
* Serato also has both realtime stems and precomputed/prepared stems, even if this is less explicitly exposed in the UI.
Here is the important point:
On Rekordbox:
* “Lower quality” stems are not really good enough for professional stage use.
* “Higher quality” stems sound better, but they are extremely heavy because they are always computed from the beginning of the track and cannot realistically be prepared for a large library unless using a very powerful gaming-class machine.
On Serato:
* realtime stems are actually very usable on stage.
* They are not as good as VDJ Stems 2.0, but they are absolutely sufficient for live performance.
* Most importantly: they work WITHOUT audio glitches on my system.
Now my actual issue:
On VirtualDJ, Stems 2.0 sound excellent. However, they require very long preprocessing on my system because I do not own an RTX 4070+ / high-end eGPU setup. Therefore, I cannot realistically precompute my entire music library.
So naturally, I expected Stems 1.0 (“Lower Quality”) to behave similarly to Serato realtime stems:
* lower quality,
* but usable live,
* and generated in realtime without audio interruptions.
But this is NOT what happens.
On my system:
* when VirtualDJ generates Stems 1.0 in realtime for a track that has not yet been analyzed,
* loading that track produces audible glitches,
* and sometimes even audible audio cuts/dropouts.
This is the core issue.
And to be very clear:
this is NOT a general hardware instability problem.
Because:
* With Serato:
* realtime stems work correctly,
* no glitches,
* quality is acceptable for stage use.
* The ONLY time Serato glitches is when my laptop is unplugged and Windows switches to aggressive power-saving behavior affecting USB/CPU scheduling.
But when plugged in and in maximum performance mode:
* Serato realtime stems = stable.
* VirtualDJ realtime Stems 1.0 = audible glitches/dropouts.
So from my perspective:
* this is not simply “expected CPU load”,
* this appears to be a specific realtime stem processing or audio scheduling issue inside VirtualDJ.
That is the issue I am trying to report.
Posted Mon 18 May 26 @ 8:14 pm
I think you may have misunderstood how Stems 1.0 and 2.0 came into existence
And most importantly, they both allow for realtime use (I and I'm sure, others as well, use Stems 2.0 in realtime all the time), and give the option for prepared use as you see fit, or of the machine is not powerful enough to do the realtime stems.
So this:
Is completely wrong.
Stems 1.0 has artifacts due to its quality level, so make sure what you are calling a glitch is really a glitch, and not expected artifacts from the lesser quality separation process.
However you can use Stems 2.0 in realtime and it was designed to be used in realtime - make no mistake on that.
- They did not arive at the same time
Stems 1.0 was the initial quality, and it had no version. Serato had no implementation of Stems at this time. - Stems 2.0 came out as an improvement of the quality
I believe that Serato's implementation of Stems came out after this update and it is the 2.0 version that you should be comparing Serato's implementation to, not v1.0.
And most importantly, they both allow for realtime use (I and I'm sure, others as well, use Stems 2.0 in realtime all the time), and give the option for prepared use as you see fit, or of the machine is not powerful enough to do the realtime stems.
So this:
Stephen kay wrote :
Stems 2.0 are designed to be precomputed/analyzed in advance. They are not intended for realtime generation during performance.
Stems 1.0 can be generated in realtime.
Stems 2.0 are designed to be precomputed/analyzed in advance. They are not intended for realtime generation during performance.
Stems 1.0 can be generated in realtime.
Is completely wrong.
Stems 1.0 has artifacts due to its quality level, so make sure what you are calling a glitch is really a glitch, and not expected artifacts from the lesser quality separation process.
However you can use Stems 2.0 in realtime and it was designed to be used in realtime - make no mistake on that.
Posted Mon 18 May 26 @ 8:40 pm
Yes, seconding what VinylTouch says - and adding that you do not need an RTX4070 to use stems 2.0 in real time. System requirements for stems are given on this site.
The computer I'm typing on has a very basic GTX1650 and I can use stems 2.0 with no issue.
Yes it takes a bit longer, but it's still usable. No need to pre-render and save to the drive.
My Alienware laptop is slightly better endowed (RTX3070), but again doesn't have a top end GPU. Plenty fast enough for live use of stems 2.0 without issue.
The computer I'm typing on has a very basic GTX1650 and I can use stems 2.0 with no issue.
Yes it takes a bit longer, but it's still usable. No need to pre-render and save to the drive.
My Alienware laptop is slightly better endowed (RTX3070), but again doesn't have a top end GPU. Plenty fast enough for live use of stems 2.0 without issue.
Posted Mon 18 May 26 @ 9:37 pm
@ Stephen Kay I experienced audio stopping glitches when I used a REV 5 for a while, but only on certain computers. Every computer I tested did have a GPU (except for Mac M4) and at least half of them had audio dropouts when loading a track. I think AT/Pioneer controllers have some issue on certain computers when using VDJ.
I also think that not having a dedicated GPU may increase this issue in your case.
I also think that not having a dedicated GPU may increase this issue in your case.
Posted Wed 20 May 26 @ 3:51 am
I have Macs (M1 and M4 Pro) and use a Pioneer DDJ-FLX10 (used with the latter also SoundSwitch on the same computer + a connected remote running on an Android tablet for a wedding on Sunday for a ceremony + reception totalling 9 hrs), a DDJ-1000 (previously owned, and used realtime stems 1.0 with this too), a DJM-S11, a DJM-S9, Phase (used in HID and DVS modes), as well as a DDJ-REV5 (playing with another DJ who owned it and a Mac, and I saw another user using it with a Windows computer without issue in a DJ workshop I help out with) all with realtime stems v2.0 and I've personally never experienced a audio dropout/glitch in any of the circumstances.
I've not really exahaustively tested with Windows, but the tests I did (I've played a couple hrs with the DJM-S11 at home with my streaming laptop), did not have audio dropouts/glitches.
Those are all Pioneer controllers.
Not saying that there can't be problems with Windows configurations, but it's far more likely to be a config issue than Pioneer controllers just not working with VirtualDJ.
I've not really exahaustively tested with Windows, but the tests I did (I've played a couple hrs with the DJM-S11 at home with my streaming laptop), did not have audio dropouts/glitches.
Those are all Pioneer controllers.
Not saying that there can't be problems with Windows configurations, but it's far more likely to be a config issue than Pioneer controllers just not working with VirtualDJ.
Posted Wed 20 May 26 @ 12:03 pm





