The new VDJ 2023 release has improved stems so it definitely needs checking out for anyone that tried the previous version. Serato DJ have their own algorithm which they tweak so that will also get better. Others mainly (not all) use the OpenAI version that’s open source and used by Spleeter etc.
The latest 2023 update of VDJ has a better update of separation in stems so all the companies are working on improving all the time, kinda how the time stretching algorithm has improved over the years as the first incarnations weren’t that hot.
To have it separated in Engine DJ and exported to the unit to be loaded in away they replace the new sampler when needed, as that can run four extra tracks simultaneously on each side. I run Serato DJ on a 2012 15” MBP and it runs great. It takes a few seconds to separate the stems so people will have to let it do that before needle dropping into the song.
I think the devs here will weigh up how long they can take to do it and would it be all worth it in the end.
The hardware seems capable from the initial test I’ve done with the sampler and looking deeper into the internals it seems it’s possible if they are pre-analysed in Engine DJ first.
The big question is “would it all be worth it?”.
Some users will love it and others not. As most of the hardware is Serato DJ and VDJ compatible, I think that’s the current workaround that will exist for a few years yet.
Maybe when all Prime models are on their “plus +” variant could we see some bleeding edge features like this. This newer generation should be able to do it but the extra headroom will be needed and we all know there’s new stuff on the horizon in Engine which could change available resources.
Would I like to see it? Yes. Do I think we will? Maybe a few years down the line.
Ease off on the negativity champ. The OP raises a perfectly valid question, and seems technically adept enough to understand the basic requirements - not that they need to prove anything to a stranger on the internet.
These forums are welcoming places to explore ideas.
Interestingly, the ‘industry standard’ have just released a very expensive flagship standalone device, promoting its processing power, and haven’t included it. That poses some questions.
They do seem to shout about their processing power in their flagships (CDJ 3000) yet they don’t have many taxing features to show it off.
Loading times seem snappy but that’s about the only advantage currently. Maybe they’ll finally add some feature updates to get the best out of it although it probably won’t get stems. They won’t even add Ableton Link.
Yeah i noticed that, its one of their big selling points this time around but the players dont actually do anything different to the old ones by and large.
Phil Morse has said several times on his stream that he doesn’t think it will be coming to standalone gear due to processing issues, i don’t know whether thats just his opinion or if he has posed the questions to the industry and that is their response.
Imagining 2 tracks with stems being mixed, even pre-analysed that’s 8 separate strings running which the software has to keep in sync and also merge into a coherent sound. I can see why that would be an issue. If you think about music production software, that’s ultra intensive and its basically doing the same thing. Once you start adding effects/loops etc too.
I think a good example would be looking at how intensive multi track playback is in the Akai Force as that’s got the same insides as the Prime range (people on Facebook have even got Engine OS running on it).
It seems to handle full tracks and VST FX with ease.
I hear they are capable of much more. They aren’t breaking much of a sweat.
Real-time : Very doubtful.
Pre-processed - Possible.
They are multi core processors with 2GB of RAM.
That doesn’t sound a lot but they use it well.
I’ve had a thought… hear me out.
Engine Lighting was developed by SoundSwitch and embedded as a cooperation between sister brands.
Well what if they do it again but bring Akai on board? Stems powered by Akai.
They could develop it based on the Force architecture and could help bridge the gap between the two brands just as SoundSwitch has. Akai could deal with dev time and interface and the Engine OS team embed it within Engine.
Mind that the Akai (which I own myself, great device) is also limited to the same Rockchip RK3288 and 2GB RAM. Only the MPC Key 61 doubled RAM and storage to 4/32GB. I hope a more powerful CPU will be considered for future product iterations. Not that it’s “weak” per se, but it’s a low- to mid-level 28nm chip from 2014, and I rather have more than insufficient processing reserves.
Similar to how Serato use Pitch’n’Time and Izotope as partners for their software… yeah that could be a good idea.
Do you think its going to come full circle and go back to the Traktor style stems with 4 pre-made files of studio quality? if it becomes popular and the software is creating 4 files anyway, people might as well just be purchasing 4 for any tracks they want to do it with. Obviously the current method in Serato is easier as it works on all current files, but they need to ramp up the sound quality massively.
That RK chip is quite dated but still covers everything we’d want for stem playback (not separation) so should be a breeze.
I almost picked one up myself the other year. They were £849 at one of the UK retailers and I still regret not getting one. I had the Push years ago and liked it.
Dear all,
Please remeber all these years we were asking for sampler option and most of the forum members were saying that it is difficult because of hardware limitation. As you see with latest firmware it was possible. So lets wait!!!
I’d like to see proper stems come back. I’ve got heaps of them as I had the Kontrol S8 when it launched as I wanted to try it.
The availability of tracks killed that. Producers didn’t want people bootlegging their tracks so were reluctant to offer them but Tooltoom and Cr2 had some great tech house stuff available.
NI open sourced it and rivals didn’t want to use it while producers were hesitant. Such a shame.
Now producers know AI can separate it they may as well go full circle, throw their hands up and say “if they’re gone remix our song, we may as well sell the higher quality version in the form of a real stem package”.
Yeah, its almost like the cat is out of the bag now isnt it, producers might as well provide them so at least people are playing good quality versions of their tracks and not ones that sound compressed etc.
i follow a Youtube channel called DJDiscoCat and he does long edits of disco and soul records, he always references stems becoming available so i assume he must be getting those from somewhere.
This is true, I was also among one of those who were convinced that they will not add a sampler due to stability/hw limitations.
I have no problems admitting I was wrong.
But…did anybody test it now with the sampler? I remember some people reporting issues with 4 deck+FX playback so it might be possible that stability WAS sacrificed in favour of getting more customers/profit.
