But the major labels are still not open to the idea, and that goes for real-time stems too. They really don’t like these stem algorithms that allow you to extract stems to remix their work in any DAW or even DJ software.
Why do you think Tidal suddenly implemented a DJ extension in their offerings when stems were previously available with the basic plan? Because the major labels didn’t like it at all and demanded royalties.
So yes, maybe the NI format isn’t dead, but it will never become a standard because the major labels will never play along. The difference with stem extraction by algorithms is that they have no control over it, and that drives them crazy and makes them feel completely powerless.
Yes I know all that! I’m not saying anything about major labels getting involved. I know they won’t.
When NI announced their stems, I said it would just be too much work for the labels to get all their back catalogue multitracks mixed down into four parts and packaged up, just for DJs. It would take forever to do, would cost a fortune in man hours and be a minimal market.
I’m just saying it would be nice to have stems packages be cross compatible.
The fact is that stems, in their most common usage, are primarily used by open-format DJs. Since major labels will never release stems in the NI format, it’s essentially useless for the majority of DJs who use stems anyway.
No reason to become deorgatory. We are not purists, but pragmatists. You stated it yourself (repeatedly) that a far more powerful chip - way above the RK3588 - would be required for reasonably speedy onboard Steams analysis. For some people speedy can mean 5x, 10x or 50x, or whatever they are used to from their laptop/desktop chips. Stronger hardware costs money, and if you have to rewrite the core system because of a different architecture, plus maintaining two OS, it costs a lot more money, that all users have to pay, no matter if they need this feature (or even Stems) at all. This is just not how this business works. You can continue talking about required AI Tops and whatsoever, or you can adapt your workflow and simply take the best of both worlds - enjoy still very capable standalone capabilties, much smoother than the current gear (which also does its job more than well enough), with the bonus feature of onboard Stems analysis for the odd track you get or add last-minute, combined with your preparation that the majority of Engine (and Rekordbox/Serato/Traktor/VDJ) users do ahead anyway, just with one extra click to prepare all your Stems as well.
I neither see any drama in that workflow, nor how that would affect Pioneer‘s marketing dominance in large clubs or festival venues at all. Their target audience couldn‘t care less if they can do Stems analysis ‚on the go‘ for their gig playlists they already have prepared before, it gives them zero additional benefit. Leaving aside that nobody seriously uses Stems on a Function One sound system - lets be honest here, the bigger challenge is to get the overall quality of Stems much improved, and I am not even talking about Engine‘s stems solely here. Progress is being made, but it takes time, and will probably requiring even more powerful chips.
Would I prefer to see a 8-core chip and 8GB of RAM in the Rane? Absolutely. Would it make a significant real-usage impact during DJing, considering real-time Stems still wouldn‘t be nearly at >10x speed? No. Not at all.
That‘s why I see this features as nice icing on the cake, a cool additional feature to get via a software update, but nothing we should be such lazer-focused on to demand a whole different chip class aka Apple M5 level inside (exaggerating here, but you get what I mean).
And if you fear AT will use this opportunity in 4-5 years and build a 10x rendering Stems chip inside their next CDJ - not gonna happen, we can bet on that. They boasted about their awesome ‚multicore power chip‘ inside their CDJ3000 and that junk still couldn‘t deliever things like dual layer or new performance features, while the even more silly 3000X made the screen a bit bigger and removed LED illuminations and the SD slot.
Nothing will happen on that front. How to make AT users to change their riders? Not by AI Tops, but the essentials, like a much improved Engine desktop and library/tag features, more robust buttons (play/pause/view…), more focused marketing, less questionable early leaks, and a proper guarantee of service repair times and avaiable spare parts. And of course, a rock solid operating system (where the newer chip and 2x RAM certainly help).
These are examples of what their people are talking about on Reddit and elsewhere. And probably - but this is a much larger, strategic decission to make - if we want to add a full DJ performance mode to Engine desktop, to combine it with a $200-250 (Numark) controller, to compete against their strongest ecosystem-entry-point product, the FLX4/DDJ400. Or if we stay on the standalone course, maybe combined with a more thight partnership with Serato & Co. The System One is a great testament for that second course. It‘s just not for the mass market, due to its price and motorized platters - but 2026 has just began, so lets wait and see
It’s not necessarily older - just quad core rather than eight. There are huge quantities of ARM chips out there, made by many companies for many purposes. It could be quad core but newer than the eight core ones in the MPCs. There could be other chips in the Rane that do the stems separation better than the MPCs. We’ll have to wait and see.
There’s a reddit thread where DJ Yoshi has posted and said he’s been involved closely with developing and testing - and he said “Wait until the next gen firmware update… something big things cooking”.
I think you mainly see this as a consolation prize
I’m sure you would have been the first to applaud enthusiastically if it had a chip powerful enough to create true stems on the fly in “load & play” mode without any pre-rendering, even if the unit cost $150 more.
Rockchip Rk3566 : 4x ARM cortex A55 @1.8 Ghz .
Estimated average price : $15 USD/ ~$10 USD after bulk purchase
Mediatek Dimensity 9300 :
1 × ARM Cortex-X4 core @ 3.25 Ghz
3 × ARM Cortex-X4 core @ 2.85 Ghz
4 × ARM Cortex-A720 @ 2.0 Ghz
Estimated Average price : 180 to 200$ per unit /** ~$120-$130 USD after bulk purchase**
**
These are two chips based on ARM technology; they’re relatively similar. Adapting the code wouldn’t have required much work.
But if you’re happy to have a $10 chip in your $2500 device…
But everyone finds their own consolation.
My workflow relies on and requires on-the-fly stems because I never know which track I’m likely to play in advance from all those in my gigantic collection (I’m an open-format DJ). And I love improvisation. So pre-rendered stems aren’t suited to my workflow.
A $2500 bonus is a steep price to pay. If the only improvement in the new generation is nothing more than a device with better fluidity, RGB waveforms, and omnisource support, I might as well keep my current equipment. It’s not going to revolutionize my life or my work, and it certainly won’t change the game.
It’s your opinion, not a fact…
In reality, the majority of VDJ users choose this software precisely to take advantage of this feature and use it whenever possible if their computer allows it. Why would they deprive themselves of it? And it’s the same for Djay Pro users.
If your device analyzes your key and BPM in real time, why would you pre-analyze your tracks in Mixed In Key? The same goes for stems.
You seem to forget that it’s not just the big clubs and festivals with Function one here that are equipped with Pioneer
There are also many small clubs, bars, and mobile DJs using Pioneer. The Pioneer market isn’t limited to large clubs and festivals; it’s actually a minority market. To ignore this is to be unaware of reality.
It would probably be capable of 10x speed with a Dimensity 9300 = 18 seconds for a 3-minute track, which is already more than acceptable for a standalone product.
Even 7x speed (25 seconds for a 3-minute track) could still be acceptable for a first-generation standalone live stem.
We remember a time when some DJ players took almost as long just to load a CD track.
It’s still better than having to wait 6 minutes to pre render a 3-minute track.
That’s a gross exaggeration; you don’t need an Apple, an M1, or an M5 to do this job, just like you don’t need an Audi RS to go get bread. There are other, sufficiently powerful alternatives at a reasonable price (see the beginning of this post in case you missed it).
Hmm, in the meantime, I can already see dozens of comments all over social media from people who think/are convinced they’ll get live stems because of InMusic’s ambiguous communication.
I’m already relishing the prospect of a flood of people complaining on the forum that – “Oh, but you said live stems were coming! I bought this product specifically thinking I’d get them, and I’m disappointed. It’s a scandal and false advertising!” – in less than six months. Just like in the good old days of Prime 4+.
I’m already getting the popcorn ready; it’s going to be crunchy.
Definitely this. I’m one of them, and there are regular posts on the VDJ forum from users wanting to know how fast the stem analysis will be with the GPU they’ve got their eye on.
Sure, but that’s no reason for InMusic to follow Alphatheta’s example, especially since any entry-level Chinese tablet has a better processor.
And if it’s an RK3568 instead of an RK3566, it’s just a $58 chip, so probably $30 to $40 per unit in bulk.
Does it seem normal to you that the processor represents only 0.4% of the product price (with an RK3566) or 1.72% of the product price (with an RK3568 bought at an average price of $40 in bulk), when it’s the brains of the device when you buy a $2500 product?
Not to me! Even if he had used a Dimensity 9300, it would have represented only 4.8% of the final product price.
It’s really only in the audio/DJ world that people are willing to pay so much for overpriced machines with such underpowered CPUs.
I’m sure that even in smart refrigerators or smart TVs, the CPU’s price represents a higher percentage of the final price.
If he had explicitly stated that this would be “pre-rendered on board” (which it certainly will be), it would avoid many misunderstandings and complaints… From there to thinking that it was deliberately made vague to boost sales by the marketing department, there is only one step…
It’s only a matter of time before we see a flood of hyper-hyped first-time buyers flooding in when they realize they’ll only get “pre-rendered on board” and not “on the fly on load”.
