Harmonic mixing has become an integral part of a lot of digital DJs arsenals, reflected by the presence of key detection algorithms in a variety of softwares. We last stood just three entrants against each other in a harmonic showdown in 2012, but this year we’ve got 12 different algorithms competing in our key detection comparison. How does Traktor Pro fare versus Traktor DJ? Is Rekordbox reliable? Is the new Mixed In Key worth an upgrade?
THE TWO TESTS
Major / Minor Triads (“Control”) – In this test, we constructed incredibly simple 30 second long songs with a drum loop and a piano playing a major or minor triad for the 24 keys. The fascinating results below show how some applications are optimized for detecting chords and others struggle with telling if they’re major or minor.
Real World Test – Instead of doing a test with music across all kinds of genres, we decided to focus on what almost each of these applications claim to do well – music that DJs are commonly playing. By that right, we took 61 of the top songs from Beatport, using all genres on the site, and ran them through our human testers and all 10 software options.
Editor’s Note: We’d really like to run a third test as another control that uses very common three chord progressions. There wasn’t time before publication but it may be added here soon.
Common errors that you’ll see notated in the results below include a mixup of major/minor – something that can be hard to detect even for a human at times, as well as “harmonically forgivable” analyses, which are off by 1 fifth. The latter error is less likely to cause DJs an issue in the mix, as there’s still a good chance that it will share many of the same notes as the track that you’re mixing in to/out of.
HUMAN BASELINE
We mention the old adage “trust your ears!” in almost every single article about harmonic mixing. It’s critical to not build a dependency on tools, but instead use them to augment a sense of when something works or doesn’t.
We’ve heard from some sources that even a groups of pitch-perfect classically trained musicians have a significant rate of disagreement about the key of certain songs.
With that in mind, we’ve only used songs in our test that had a clear consensus (read: simple majority) from our human test group. Our test group included myself and 6 awesome well-suited volunteers with experienced ears (even a few perfect-pitch ears in the group) keying tracks from the DJTT community – thanks for listening to these tracks and helping out! It is possible that our ears are somewhat fallible, which is why we’re putting all of our data at the end of the article. If you disagree, feel free to make a case for a different consensus on a track in the comments.
THE COMPETITORS
Beatport’s Key Detection
Beatport has their own detected key for almost every song in their catalog – which is why we started with their own catalog as a place to source tracks. Artists aren’t given the option to report their own keys to Beatport – so even if the author has an associated key that they know they wrote the song in, it apparently gets tossed out and replaced with a detected key. The algorithm has always needed some work (“We’re still refining our BPM and Key detection” reads their support page), but if an avid fan or angry artist spots an error, they can report incorrect keys and BPMs.
Major/Minor Triad Test: Not possible – no way to manually test.
# of correct tracks (vs human volunteer consensus): 40 (66% correct)
# of Major / Minor misreads: 10
# of results off by a single fifth: 7
Mixed In Key 6.0.2
This is the most recent version of Mixed In Key’s algorithm, having first made a début hiding inside of the company’s DJ software, Flow (read our review here). The update was released at the end of December 2012, with a number of changes to the algorithm. At the time, MIK’s founder Yakov noted that the changes were significant, “the old version would group the bassline together with the lead synths, the pads, and so on. The new version can “hear” music in higher resolution, and analyzes each track in 6.83x times more detail (over 600%) because it listens to the entire piano roll.”
The 6.0 version of Mixed In Key costs $58 for new users who have never purchased a copy before, and $29.99 for users of 5.0 or earlier – and perhaps worth the money, considering the impressive results below.
Major/Minor Triad Test: Perfect 24 out of 24 – every single key correct.
# of correctly analyzed tracks (vs human consensus): 58 (95%)
# of Major / Minor Misreads: 0
# of results off by a single Fifth: 2
Mixed In Key 5.6
We make it a point to include the previous version of Mixed In Key here as a simple point of comparison – most people who own the software likely don’t own the most recent version and want to know if it’s worth updating to. This version of MIK likely performed better in this roundup than the 2012 one because this test included mostly DJ-oriented tracks.
Major/Minor Triad Test: 20 / 24 correct.
# of correctly analyzed tracks (vs human consensus): 55 (90%)
# of Major / Minor Misreads: 0
# of results off by a single Fifth: 4
Algoriddim Djay 4.2.2 (Mac)
Algoriddim’s software has been around for a while, but found major success on the iPad. Unfortunately the iOS version doesn’t have key detection, so harmonic mixing and testing the most common use of Djay isn’t possible. Instead we’ve tested it with the Mac desktop version.
Despite doing well on our rudimentary major/minor triad test, Djay struggled to have a precise analysis of the keys – often being off by just one fifth. It’ll do in a pinch, but serious harmonic mixers might need something more accurate.
Major/Minor Triad Test: 23/24
# of correctly analyzed tracks (vs human consensus): 42 (69%)
# of Major / Minor Misreads: 4
# of results off by a single Fifth: 12
Pioneer Rekordbox 2.2.0
Pioneer’s library analysis and organization software is where most CDJ-based DJs go for setting cue points, making playlists, burning CDs, and compiling flash drives for their sets. It also has built-in key detection, which makes it yet another candidate for our roundup!
One thing we found about Rekordbox was that it was painfully slow compared to all the other applications. Even loading in a single track for analysis brought the app to a crawl, making it difficult to do anything else while waiting for the results – which as it turns out aren’t that amazingly accurate. Perhaps you get what you pay for, as the Rekordbox application is a free download.
Major/Minor Triad Test: 21 / 24
# of correctly analyzed tracks (vs human consensus): 35 (57%)
# of Major / Minor Misreads: 9
# of results off by a single Fifth: 9
Keyfinder 1.25
Keyfinder is the free open source answer to the paid solutions for key detection on the market. It’s cross platform, and is capable of analyzing a massive number of audio codecs and can write to metadata tags just like any larger paid option. Interestingly, the software is capable of doing a detailed analysis that displays a “chromagram” of the notes in a track and assigns a primary key for different sections of a track (see image above). Most users aren’t looking for this level of analysis – but for a completely free software, it’s probably the best value of any of the programs in this test.
Major/Minor Triad Test: 18 / 24
# of correctly analyzed tracks (vs human consensus): 47/61 (77%)
# of Major / Minor Misreads: 3
# of results off by a single Fifth: 7
Traktor Pro 2.6.6
Traktor is another recent entry to the key analysis game, having implemented key detection in Traktor Pro 2.6.1 almost one year ago. The key detection algorithm originally was shown to be roughly equal to Mixed In Key 5’s in a casual test done by the developer behind Keyfinder. In reality, we found that the software wasn’t so good at our short sample major/minor triad tracks – but absolutely crushed when it came to real world track examples.
Update: We’ve run the Traktor Pro test a third time with a brand new set of these audio files – apparently stripping ID3 tags isn’t enough to remove all of the key information from Mixed In Key. We’ve revised our results – which are actually nearly identical with Traktor DJ on iPad.
Major/Minor Triad Test: 11/24 – perhaps it requires more context than our simple tracks?
# of correctly analyzed tracks (vs human consensus): 33 (54%)
# of Major / Minor Misreads: 9
# of results off by a single Fifth: 11
Virtual DJ Pro 7.4
While we continue to wait for updates on the ever-forthcoming Virtual DJ 8 (potentially next week at NAMM), we’re forced to use the available software to judge Atomix Productions’ key detection abilities. We’re betting that VDJ8 will have an updated algorithm – and will likely do a bit better at telling major and minor triads apart than VDJ7.
Major/Minor Triad Test: 12/24
# of correctly analyzed tracks (vs humans): 40 (65%)
# of Major / Minor Misreads: 1
# of results off by a single Fifth: 14
beaTunes 3.5.17
BeaTunes 3 has been out since 2011, but since it received an update in November to the most recent version, it remains on our list of software to use for key detection and analysis. The application can actually be used for a lot of very useful automated library cleanup work in addition to key and BPM analysis – including fixing incorrectly tagged file names and ID3 tags. That being said, the key detection algorithm needs a bit of work for it to be at all reliable for DJing. The software is available for Windows and Mac OS X for $35.
Major/Minor Triad Test: 23/24
# of correctly analyzed tracks (vs humans): 27/61 (44%)
# of Major / Minor Misreads: 8
# of results off by a single Fifth: 16
Rapid Evolution 3 Beta 58
Unfortunately Rapid Evolution hasn’t seen any development since our last roundup in 2012 – but we know there are some users out there who are loyal to it still, so we included it here. The software doesn’t fair that well in the outcome.
Major/Minor Triad Test: 19/24
# of correctly analyzed tracks (vs humans): 36 (59%)
# of Major / Minor Misreads: 9
# of results off by a single Fifth: 7
Traktor DJ 1.3.3 for iPad
At the request of a few of our readers upon finding out that we were doing another key detection showdown, we’ve separated out Traktor DJ on iOS from the desktop app.
Update: Just two of the keys detected differ – meaning that there isn’t that much of an inconsistency versus Traktor Pro. Have a look at the numbers:
# of results different from Traktor Pro: 2 more incorrect
Major/Minor Triad Test: N/A – Traktor DJ can’t view our 30 second test files. We’re investigating.
# of correctly analyzed tracks (vs humans): 31 (51%)
# of Major / Minor Misreads: 10
# of results off by a single Fifth: 11
CrossDJ 1.6.1 for iPad
CrossDJ earns the title of being the other iOS application to make the list. Not only has the software expanded to the Android platform, but it continues to see numerous updates and support with regularity from a dedicated development team. Worth noting is that CrossDJ is developed by MixVibes, who are also the team behind Pioneer’s Rekordbox software.
# of results different from Rekordbox: 26 – clearly not the same algorithm at all.
Major/Minor Triad Test:
# of correctly analyzed tracks (vs humans): 48 (79%)
# of Major / Minor Misreads: 3
# of results off by a single Fifth: 7
THE DATA FOR YOUR INSPECTION
Each test was run twice to make sure there were no testing discrepancies – and we also eliminated any songs where there was no clear consensus among our human listeners.
You can access the full Google spreadsheet here – there’s a lot of interesting info in here that could be extrapolated, and even links to all the songs on Beatport if you want to listen in.
Want to dispute a key that we chose as the “correct” key for a song? As we noted earlier in the article, there’s always going to be a bit of disagreement on certain tracks – which is why we ran a control test.
One last note from the editor: this article has in no way been sponsored, endorsed, or even asked for by any of the involved software developers. It was fueled entirely by fruit snacks and hours of clicking, listening, and analyzing.
[…] con beatunes o con Traktor nonostante divesi software hanno lo stesso sistema montato. In questo articolo in inglese vengono messi a confronto i maggiori software di rilevazione delle chiavi armoniche […]
[…] aussi, le taux de fiabilité reste très discutable. Selon cet article, la plupart de ces logiciels ne dépassent pas 70% de taux de réussite pour trouver la bonne […]
[…] face is the home windows 7 auto shutdown problem. If you are you looking for more in regards to Detection look into the web-page. Inexperienced tea can be a typical ingredient; analysis from the January […]
This is an old post… how this algorithms stand in 2017?
Please include Queen Mary key detector used in Mixxx
[…] http://djtechtools.com/2014/01/14/key-detection-software-comparison-2014-edition/ […]
[…] http://djtechtools.com/2014/01/14/key-detection-software-comparison-2014-edition/ […]
It would be interesting to see where Serato’s new detection falls.
Are you guys doing a 2015 Edition? With Serato 1.8 coming out end of this month, a lot of people would find it interesting to see how SDJ performs against it’s competition..
[…] Key Detection Software Comparison: … – Harmonic mixing has become an integral part of a lot of digital DJs arsenals, reflected by the presence of key detection algorithms in a variety of softwares. […]
It’s been a year and half, there might be upgrades of accuracy in each softwares, reckon DJ Tech Tools should be doing a new review on key detection softwares?
Just saying 😉
BUMP ! (for justice)
[…] ??????????? ??????, ?????????? ????? ???????????. ?? ??????????? ?????? ?? 2014 ???, ???????? ?????? ?????, ??????? ???? ??????? […]
If you want to reorganise your library by complementing filenames or IDv3tags using detected key’s and BPMs from VirtualDJ or rekordbox database, check my program 😉
https://sourceforge.net/projects/vdjrenamingtool/
I want to ask for your opinions 🙂 I usually play jungle, ragga. I want to know which key detection is more accurate: Virtual DJ 8’s or Key Finder 1.25’s. I have made a few screenshots of my Virtual DJ 8 (and joined them together). On the harmonic tab you see the key detected by Virtual DJ and in the comment section (the last one) the key detected by Key Finder. Which one is more accurate. I know I have to listen, but It will help to know which one is more accurate.
Also I think the price of Mixed in Key is way to high and the fact you have to pay for updates is just ridiculous ….
I think Traktor should buy MIK – that would be a great integration of technology right there.
Keyfinder is the “VLC” of key detection.. Impossible to don’t support that quality soft..maded with passion and talent. Thanks again to share this with us.
[…] Key Detection Software Comparison 2014 – DJ Tech Tools compares all the key detection programs out there to conclude as ever that none is close to prefect. Moral: always trust your ears, folks… Read more […]
[…] Marketing Professional Rekordbox does a bit better than Traktor….Mixed is Key is by far the best.http://www.djtechtools.com/2014/…Embed QuoteUpvote • Comment Loading… • Written just now Add your answer, or answer […]
[…] tunnistusalgoritmit eivät ole täydellisiä ja niiden laatu vaihtelee ohjelmasta toiseen. Käytän itse Traktorin oman sävellajitunnistuksen lisäksi ilmaista avoimen lähdekoodin […]
nice article, ive used mixed in key since version 3 and now have version 6. I just bought a bunch of track from beatport and was trying to decide whether i should overwrite all the beatport info ie key and bpm, with MIk 6 or not? according to this article, the best bet would be to just trust mixed in key i guess.
I have to wonder in this day and age why it isnt industry standard for all producers to include key and bpm in the title of the tracks they make…wouldnt that make things a lot easier for everyone?
Hi all,
I did not hear all the sound but let me tell you that you’re completely wrong about the song Almost Home (Agoria Remix) that the human consensus hears as a F# min (assuming that most of the tools are wrong) while it is actually sometimes A min (or 10A), sometimes D MAJ (or 10B).
Now, if only Denon’s “Engine” software wasn’t complete garbage. I never thought I’d find something worse to use than Rekordbox but here we are.
I’m sorry i missing something how about BPM
but i like the test thanks
Funny thing here … to me, a software developer by day, crime fighter by night – and a DJ whenever it fits in the schedule … at one point (right around when Traktor introduced key detection) Mixed-In-Key and Traktor were using the same detection library (zplane’s tONaRT) … but a quick look at zplane’s site seems to assert that this is no longer the case?
The problem with this is of course – if you change your detection algorithm at some point, every key detection before the new one becomes “invalid” (or actually, “inreliable”) with everything detected with the new algorithm … ie. they should no longer be concidered “compatible”. Can’t help but wonder what NI is using these days?
Hey Markku, here’s info direct from the source. Mixed In Key used tONaRT library in Mixed In Key 1.0, 2.0, and 3.0 which we released back in 2008. We created custom technology on top of tONaRT for Mixed In Key 4.0 (which gave better results), and made a huge effort to improve our results for Mixed In Key 5.0 which used mostly custom technology with a tiny bit of tONaRT underneath it all. We haven’t used “stock” tONaRT in 4+ years, because we wanted to create our own algorithms.
The new Mixed In Key 6.0 uses 100% custom technology and no longer uses tONaRT at all.
Interestingly, zplane site still lists MiK as “product that uses their library” whereas Traktor no longer is on that list. Granted, whether software package even wishes to be listed or not probably depends on the options of their license.
For some reason they are still listing MiK there… of course, in software development, to “build something on top of something else” strongly implies that the “something else” is still the low level foundation … somewhere down there. I do not imply to know any specifics (because I don’t) … just reading some marketing blurps 😉
We have huge amounts of respect for zplane, and some of our software (iMashup, Mashup) still use parts of tONaRT. So, their listing is accurate and the quality of their technology is fantastic. Their Elastique code is the industry standard for beatmatching music inside software.
Heh… some times I really feel like I’m not running on all thrusters. Finally connected the dots of who it is I’m talking here with… had a funnny feeling “this nick feels kind of familiar from somewhere” 🙂 Could’ve just taken your word as gospel from the get go 😀
You didn’t account for relative minor/major scales. According to your chart Rekordbox recognised Alan Fitzpatrick’s ‘ We Are Forever Young’ as C#M, not A#m. C# major should have the same notes as A# minor, providing there is no accidentals.
Mixxx lead developer here. Great writeup! I would have loved to see how the Queen Mary University key detector stacks up.
http://vamp-plugins.org/plugin-doc/qm-vamp-plugins.html#qm-keydetector
It’s also open source and available as a VAMP plugin. Mixxx is adding key detection in Mixxx 1.12.0 and will have the option of using KeyFinder or the QM key detector. You could try out the QM detector in the latest alpha releases of Mixxx 1.12.0.
Feel free to email me if you’d like to try it out: rryan@mixxx.org
I second this, the only thing lacking currently is a program that can write the result to ID3 tags.
Can’t wait for mixxx 1.12
Has this test happened yet?
Kudos to Mixed In Key for achieving the 95% accuracy rate reflected in this report. This rivals the accuracy of professional musicians used to compile our online database at camelotsound.com
the best key finder i own are my own ears! the software is great for organizing music. but part of learning my music is knowing which keys work with what.
Thanks for this great comparison. I’ve been on the harmonic mixing trains for years now, and I like to see how this topic has progressed over the years.
That said, thanks for finally proving a long term theory of mine, that is that VirtualDJ does not care about MIN/MAJ detection. My own observation showed that it always detects minor keys. But of course I could not be sure if this is a subjective impression of mine or not. But with the Baseline results shown in the google doc it is pretty obvious. 🙂
Djay 2 for the iPad has key detection, but requires a 64-bit device (iPad Air).
Kudos to your starting the new year out with such a good article.
Absolutely agree on getting DJ Endo’s opinion on this. Also a follow-up with Native Instruments on their poor results would be interesting.
Here is a chart that compares the results from DJTT and the one by DJ Endo from August 2013 (only taking the minor keys, as DJTT tracks are 95% minor keys)
Heh, I admire your use of spreadsheets. I used to use these frequently to show why my ideas were good ones.
Really!!! No Melodyne analysis?
I am an avid strong supporter of BeaTunes and have used it for years. This app is priceless to me in adding key, color, missing data, year, genre and most recently album art. The Matchlist function is the bomb. Yeah, some DJ’s are gonna whine and cry “know you’re music” blah blah, but look; I belong to a record pool and get hundreds of new songs a month. I cannot “know” every new tune out there and keep things fresh. The Matchlist is a tool, not a crutch and it helps me tremendously. If you have a very large library and use iTunes, when you initially get the program it will take time to sort thru. But once you’re data is synchronized, adding new songs and keeping things current is very easy. Well worth the money.
When MIK 6 was released I though why should I. All my DJ software have key detection. Now I see why and I’ve upgraded 😀 . It looks like MIK has some problem with DDJ SX ass it won’t analyze any song while it’s connected
Do a training with your ears and after a few weeks you have no problem finding the keys with a keyboard.
for harmonic mixing, just make your setlist and search the keys. The harmonic mix shouldn’t be the most important thing during a gig so it’s just a nice side effect while mixing and if you have the keys then you can mix sometimes after the camelot wheel
I agree that that’s a great skill to have – and it’s important to check your keys on any sets that you have planned out. That being said, many DJs don’t just “make a setlist”. Being able to have the key of any track in your library is something that only automation or incredible amounts of free time will accomplish.
Thank you for such an informative test!
Mixed in Key is $58, Flow is $58 but contains Mixed in Key as part of the software. Is there any reason to buy MIK over Flow? Is the Flow analysis data not as portable or something?
Flow is a self-contained app – it doesn’t update ID3 tags, so the results are only visible inside the software itself. Mixed In Key is like a plug-in for Traktor, Serato, Ableton Live, rekordbox and iTunes.
The question which one you should get depends on whether you want to mix 100% inside Flow, or whether you want to use other DJ software
aha makes sense! thanks
I had huge hopes on keyfinder, Ibrahim where are you man 😀
I’m on it! In the next few weeks I’ll be starting a crowdfunding drive to get 1000+ songs analysed; that should allow me to up Keyfinder’s accuracy and will give us as a community a big database with which to do comparative tests like this one.
Thanks for your support =)
keep it up Ibrahim!
After doing my own comparative test of 100 tracks of Keyfinder vs MIK, Keyfinder was the most accurate at finding the correct key with correct scale. I mainly use MIK just for BPM tagging. Im a big fan of Keyfinder, since I do live p.a./ hybrid dj sets, being in the correct key is crucial and Keyfinder does that job to near perfection.
Yep, same for me. I started to blindly trust KeyFinder by now. The accuracy is so high for me, that the few misdetects that I might find by keying manually are not worth the time invest.
But I would also love to see an update of KF based on this test. Not sure if the dataset present in this test is big enough for Ibrahim to finetune his algorithm further though.
And the winner is …. of course. That’s why I been relying on MIK since version 4.
What about plug-ins? Anything work for using it to detect samples in production such as ableton?
Ableton Live doesn’t show any meta-data, so it’s hard to browse samples by key. I’m a Live user, and here’s what I usually do:
1) Select all the files in my Basslines folder on disk, and drag them into Mixed In Key. Analyze them, and save them to a “Basslines” playlist.
2) Do the same for all other folders I have. The result should be a separate playlist for each type of sound – Bassline, Pads, Leads, etc.
3) Drag and drop from Mixed In Key into my Live session
It’s not elegant but it works -Yakov
I think Traktor Pro 2.6.6 accuracy is wrong. I just bought 13 of these tracks from Beatport, and ran them through Traktor 2.6.1 and Traktor 2.6.6. Both of them gave the same results as Traktor DJ, so NOT what’s shown in DJTechTool’s Google Doc.
I think Native Instruments is reading the Mixed In Key 6 tag from the file and showing it as their own. Dan, did you analyze in Traktor *after* you analyzed in Mixed In Key?
All of the files had their Key tags stripped from the ID3 tags after each test. That being said, I’m about to run Traktor again as we also were surprised as this outcome and discrepancy between Traktor Pro + Traktor DJ
Forgive me please, but good methodology would have removed this potential for error. Put all the songs in a directory, make a copy of that dir, do key analysis on the copy, repeat for each app. It’s the only way to prevent previous runs from contaminating your data.
ever think about testing mixmeister?
the current possibilities are still a bit skimpy.
so you still need to feel it, – in the egg * (political correctness) or in the breasts
pure intensities and the themes of the song will be for ever made by hand
I had an idea this would be the result being an avid Traktor user, I can get songs mid set, throw them in and not once have I ever been let down by Traktor’s harmonic detection. Very impressive to see MIK has upped it’s game to a similar level though, I remember using KeyFinder because it was the best AND free.
thank you for this artice! it saves me about 60 $… or about 40 €.
i have thought about buying Mixed in Key 6 about two weeks ago, but if Traktor (i own it since 4th December…) is more accurate than their key-detecting-Software, i won’t buy Mixed in Key.
best regards from Germany and have a nice day!
I’ve updated the article above with some more recent results – Traktor is actually just really good at finding some metadata stored by Mixed In Key
Did you clear the key Tag in your file before the test was done?
Because Traktor Pro use previous enterer key tag during there analyses,
in the result will be very similar then previous Key tool.
On cleared Tag the result will be more different.
See also: http://blog.dubspot.com/endo-harmonic-mixing-key-detection-analysis/
I think you’re right. Traktor may have been reading ID3 tags during this test, since Mixed In Key was used first. I am not getting the same Key Results for Traktor that Dan posted in his Google Doc
That’s the interesting thing – the ID3 tags were cleared between tests, but it seems like Traktor still managed to get its hands on MIKs results. The article is updated to reflect.
Bummer! I’m sure many of us would have hoped that the Traktor results would be really that good. But 54% is very disappointing.
At least it’s clear now: if you want harmonic mixing – then either you can have half-decent results with Keyfinder or you need to invest 58 $ for Mixed in Key (which is what I will do).
Thanks again for the test and for updating the article following the comments on Traktor.
Yeah that is disappointing, I was hoping Traktor would have been waaaay better!
I do like their colour system though.
Colours are pretty =)
After you run your tracks through MIK, Traktor’s colors are correct. Their 1m is in A minor and is the same as 8A on the Camelot wheel. MIK doesn’t just add the Camelot tag, it writes over all the old, incorrect ID3 info for key.
Traktor – 98% ?!?
That’s pretty damn accurate!
Thanks for the very interesting test
Most striking for me is that Beatport has wrong key information for a third of their songs – and even more: artists don’t seem to care about it
btw: artists play her style of mood. There is no recipe for this.
most of the studios have her own style. And all of them her individual sight.
I’m pretty sure Cross on the iPad uses my Keyfinder algorithm, which explains how close they are. Looks like I have some work to do on accuracy, though! Thanks for including KF, Dan!
You’re welcome – interesting that Cross uses Keyfinder. Does Rekordbox not use it?
Kinda bummed out Rapid Evolution 3 isn’t tested : would’ve loved to see more free alternatives other than KeyFinder
It’s in there! After beaTunes and before Traktor DJ…