thats a good idea Despo, unfortunately DJay beatgrids wont transfer to any pther DJ apps including Serato, you can transfer cue points,

The way DJay grids work is very different to Serato Grids. It has some form of A.i.

I think DJ Grids info is stored onto DJays saved library file that only DJ can read.

Okay so according to Mix Master G, developer of DJCU, Djay pro generates it's beatgrids in real time using ai, which is why this data cannot be converted to serato.

Damn ittttttt.

Serato get a move on and develop fluid beat grids, I'm tired of gridding

lol, I bought Mix Master Gs application and I cant get it to run,what a waste of money.



I agree 100%

Besides what was stated already (DJay doesn't store beatgrids conventionally, it generates it on the fly with AI), you'd also find that even if you could convert the grids, it may not be placed correctly depending on the file type. mp3 files could be interpreted differently by different software (an encoder issue), causing unexpected beatgrid and cue shifts, and you'll essentially have to correct the beatgrid manually again (defeating the whole purpose). It would probably work well for other encodings though, like FLAC, ALAC, WAV, etc.

Precise fluid beatgridding is a major deal Serato will need to put research in to figure out. I think it could be based on Drum Stems with a built in metronome click

It is intriguing that Serato possesses the technological capability to create flexible beatgrids. Serato Studio has incorporated this feature since its initial releases, yet the reason for its absence in Serato DJ remains unclear. Consequently, my current workflow involves analyzing all tracks within Serato Studio. Given that both applications share a common database, upon returning to Serato DJ, the grids exhibit red markers (anchors) at intervals of 1, 2, or 4 bars. Please note: it is essential to close Serato DJ prior to launching Serato Studio to facilitate track analysis within the latter

this tech could be patended and copyrighted by Djay

On the fly generation might be patented. Beatgridding itself through regular upfrontanalysis probably isn't (all recent software has it), and this can be inplemented as an upgrade to existing upfront analysis using AI (not on the fly).

bare in mind DJay has the backing of Apple and their software developers are on another level,



seems like on the fly beatgrids is the future, upfront analisys is the old way cause it requires you to adjust beatgrids then save them, I can quickly make live changes while DJay is playing a track. I love it.

now with the new update it shows the BPM changes in the track.

I don't think it matters tbh...you can achieve the same thing up front, it would do the same thing but write the entire result permanently to the database/file afterwards - you can always edit it after if required if it's not right. On the fly is more nuanced...besides not being to edit a saved version of it (by definition), in particular, you can't share it (a way of "locking you in" to the ecosystem) - which means third party tools that depend on sharing in the conventional way won't work, at least without rethought on how to share it (think SoundSwitch and DJCU).

Remember, the problem wasn't how you generate the beatgrid, it was the accuracy of what was generated in non-uniform BPM cases.

Serato needs to implement this feature in 2025, otherwise I'm out. I'm pretty sure this is a feature many user would actually be willing to pay for.

Is this a fact? This would seem like a very strange design choice to me, what happens when you manually adjust a beat grid in djay?

DJay seems to be a hybrid - it most likely does auto detect a baseline beatgrid, but the system still can modify it in realtime based on fluctuations it finds in ranges of the song as it plays. The edits you make (adding anchor points, etc) would also be stored but also probably results in recalibration of the AI system from those points onwards. This is different from conventional gridding, which is all upfront, and unchanging - the "dynamic" part wouldn't be convertable by a tool like DJCU after the fact.

Why was this choice made? Unclear really - it works but it could be done completely statically/upfront if necessary (a song doesn't change between plays). However, not being convertible by regular tools does work in favor of keeping you using DJay, so at least from that perspective, the complexity makes sense.

your right in my experience, I dont think its on the fly cause, DJay grid is not 100% perfect,
if it was on the fly than DJay will change any corrections you've made to the track, and it will be frustrating.

Ive had plenty of songs DJay got it wrong, I've had to adjust multiple downbeats,

Even A.I cant tell the downbeat of a song,

once DJay analyzes a song thats it, it wont touch the grids unless you analyse the track again,

so I can go in and manually grid edit the track with the correct downbeat, then this info and grid will not be touched by DJay unless I reanilise that same track.

also BPMs are always off, the track is 160pm but it thinks its 80bpm.

this is why I dont think its on the fly analysation.

I think AI is innitially used to analyse tracks the first time Thats it. not on the fly.

I also think this is a patented tech by Algoridem, so Serato will have to come up with something else. thats different but the same concept.

Thanks for clarifying. A bit strange that people just make some claims here without really knowing better. I also watched the Mixmaster G video in the meantime, and this conclusion would be a misinterpretation of what he is saying.

Why any software would waste resources by continuously keep rerunning an algorithm it has already done is beyond me.

Too bad, I actually just started a djay trial last night in hopes of being able to transfer the grids. I'm sure it would be theoretically possible to transfer the beatgrids, but at the same time I'm sure Algoriddim has no interest in this being done so they will obfuscate the data.

I have a feeling Algoridem is purposely making it impossible to transfer grid information over,
just like how its hard to transfer cuepoints and crates over to DJay.

I did use Mixmaster Gs software but I just gave up,
The more you use DJay the more you will fgure things out cause there really isnt any tutorials to do the things you want out there on the web.

Im been playing around for almost 8months now and I've just self taught myself on how I want Djay to work for me especially mapping my SX2 to do exactly waht I want on DJay including full custom mappings for Stems including Stems Efx more then what you get from Serato.

I'm really happy with the results.

Also when DJay brings out updates, It actually brings out new features and not just an update for new controllers and DJ hardwares.



I agree, You have a lot more knowledge if you actually use DJay fulltime.



Yer I agree it totally makes sense,

I think it's very easy to transfer the crates to Djay pro from internal or external drive without any other software.

On Mac (also i believe in windows laptop), just run "UI Configuration Assistant..." and you will see that it found the crates.
Select the crates folder location and that's it.
My external thunderbolt ssd takes about 4 hours to complete on M1 Macbook Pro for 65,000 files.
Then my crates are in my collection with all my subfolders.

I also transfer the library file that was created to my Ipad Pro M4 and I have all the crates in Ipad.

Also all my cue points are there because Serato writes them in song tags

I use the inbuilt DJay crate import tool for my external and it works fine, the issue I have is everytime I update my library I have to manualy update the new files, it doesnt do it automatically, this is my main issue with DJay library.

i have over 100k files so it takes a full day or more just to transfer my crates.

How do you think stem analysis works then in most DJ software currently (and most DJ software gives you an option to prepare stems as secondary)?

In all analysis cases the goal is to learn as much about the music so that you can DJ with with as little overhead/resources/time required.

In that case (stems), the desired output is separate audio tracks (vocals, drums, etc) to be selectable and used for mixing...storing that would balloon you collection or database size (leaving less space for your music/other cillection metadata). Modern day computers can have enough RAM to store such for up to 4 decks at any instant, so it would be more space efficient (at the expense of time/CPU/GPU resources) at song load/initial play. I guess my summary is - repeat work for analysis is already being done by the software you use and there is a reason for that.

With that said, this is why I suggested what I said:

The upgrade to beatgridding has a desired output of better placed beatgrid markers and would naturally benefit from already existing AI tech (proper anchor placement is nice to have but not a core requirement as more "awareness" of the music is required = more training/processing). The most accurate AI usage we know of to date is the upgraded stems usage, but that is pretty intensive (timewise and resource wise) to do completely upfront without users noticing (especially in batch analysis). However, DJay's analysis is still pretty quick. So what I think is happening is they are using reduced quality analysis to generate the grid (as smallest time for analysis is important), but when the track is loaded, they use the best analysis to fine tune. However it still respects your edits if you make them - it just means in depth analysis resets itself from that point for fine tuning, respecting the changes.

Side note: 1) I also attend most of MixMasterG's streams when he does them live - I attended his DJay special as well l...I wasn't referring to or making any claims based on his video. 2) I'm not claiming to know what Algoriddim have done - I'm just speculated based on how I've seen it perform.

I'm not saying they couldn't use obfuscation here to promote lockin, but I don't know if that's in their best interest long term (proper SoundSwitch integration would probably be a larger undertaking for e.g.), and MixMasterG and others do have a relationship with Algoriddim - they would have most likely worked out the details by now as they would want migration to their product (which is what DJCU provides), but it could be a patent/licence issue as well here too if the details.

DJay pro is so far the only one to have this feature at this moment as far as I know, not even virtual DJ has it and they're mostly implementing cutting edge technology into their stuff, even were the first to implement AI stem live seperation into their software.

serato is probably going to get it at some point, most likely after virtual DJ does, but it would be great if someone developed a third party AI tool to edit serato beatgrids like mixed in key does for keys.

Not to hijack this thread too much, but I'm a bit confused as to how the gridding in djay actually works.

Example 1: Prince - I Wanna Be Your Lover

I gridded this song manually on Serato and came up with a tempo from 113-119 bpm.

The same track on djay plays at an even 114.8

Example 2: Earth Wind & Fire - September

djay starts at 117.4, changes up the tempo a couple times, but after 10 seconds it's at 126.0 for the rest of the song.

I'm assuming djay is able to play songs with an uneven tempo at an even speed.

Is it possible to turn this feature off?

How does it decide whether to change the tempo within a song, when the speed change is just too big?

Could someone explain how Ableton Link is in Serato DJ Pro for 8 years but still absent in Serato Studio released 3 years after?
www.digitaldjtips.com
djworx.com
www.engadget.com

I normally don't crosspost features on competing software platforms, but given the origin of this post's originated from that, I'll just make a similar metion here:

I'm surprised to see that no one mentioned that Rekordbox now has improved beatgridding as of v7.0.7, released on December 17, 2024 (rekordbox.com) (However, I won't take credit for finding this, another colleague on another forum pointed this out):



The setting is off by default, but did seem to work well on an existing variable-bpm track I tried (Bad Bunny - Despues De La Playa).
It also seems to be done fully upfront/is fully exportable via existing library conversion tools as well (I've verified successful export/import of the beatgride using one of the major inter-library conversion modification platforms).

Hey thank you for this news, that's great! Did you successfully convert beatgrid data to serato and which platform did you use?

This is huge until serato manages to implement ai beatgridding

So I didn't do the end result trial to Serato, but 1) the platform's grid representation on import of the track showed all the variance in its grid (Lexicon was the platform), and given that is the case, it will sync to any destination software. DJCU would work just as well too in theory (I'll try that later).

just looked into this. There are beatgrid information stored in rekordbox, which look something like this:

BeatGrid {
unknown1: 0,
unknown2: 524288,
beats: [
Beat {
beat_number: 3,
tempo: 12371,
time: 30,
},
Beat {
beat_number: 4,
tempo: 12371,
time: 520,
},
Beat {
beat_number: 1,
tempo: 12371,
time: 1010,
},

etc..
This can be technically converted to the serato tag format. Dont know if I find time for this to develop. But yes it is possible to convert this dynamically analyzed beatgrid to serato

finished a proof on concept. conversion seems to work

I also confirmed existing tools can convert thw Rekordbox grid.to Serato. @SnipeStyle also hopefully your POC takes into account the infamous cue shift issue with mp3s

so there seems to be also a limitation of the amount of marker you can set in serato per track. This is not good for long or fast tracks

So what's the best tool to convert beatgrids to serato right now? And does it work reliably?

I just wrote a script for myself to test this process. I dont know a official tool. However, it will never work 100% for all tracks, due to the marker limitation per track.

Maybe someone could make a video on YouTube to explain exactly how it works , step by step.
I also notice that Rekordbox have a better algorithms for beat grids , so if it’s possible to transfer to Serato , will be awesome.

I just wrote a script for myself to test this process. I dont know a official tool. However, it will never work 100% for all tracks, due to the marker limitation per track.

So , it’s working your script ?
What’s the best tool to transfer beat grids from Rekordbox to Serato ?