AI Slop has been flooding the music industry. Itās sofa king infuriating!
This is Digital DJ Pool and just today over 30 AI Slop submissions. I contacted support at Digital DJ Pool and they 1) didnāt acknowledge it as a problem and 2) admitted not to having a strategy to mitigate.
If you are against AI slop in music, the best you can do is to NOT download or play artifacts generated by those unskilled people.
Itās sad to think that this is likely going to become the norm and will simply just take over, imagine working on an album putting countless hours into only to be surpassed by an AI side mission sad times
I think the electronic music community brings it upon theirselves, with all kinds of āinstant gratificationā features being asked, released, and greedily consumed. Both in the DJ and production side of things.
I am often irritated by the mixtapes one of Belgian national radios (MNM) airs: a lot of them feature stems and harmonic mixing, which is cool by itself, but they often just sound dissonantā¦. Seems the DJ in question must be relying on the Camelot system. But just looking at the Camelot system is no guarantee a mix will work: you can hear a lot of chords, notes, or even voice leading that just clashes harmonically, despite being in the same key. And the Camelot system only shows 2 of the 7 possible modes one diatonic scale offers. Camelot systems augments the chance something will work, but itās no guarantee! But it is ofcourse easier to just look at the key in that little column, donāt use your ears, and donāt ask questionsā¦
At the same time I often see video ads showing of production tools that automatically write your chords and harmony. Hell no, I want to learn this myself! But apparently there is a market for these toolsā¦
If DJing and producing is reduced to just matching a numerical value and subsequently pressing a single button, ofcourse AI is very close around the corner to completely take overā¦
But hey, I must be old fashioned, seeing how many people feel an instant-gratification feature like stacked waveforms should be the norm todayā¦
And apart from the artistic side of things there is also a monetary model: be it Spotify, YouTube, or any other platform: you get paid per click. Putting out quality content gets clicks, but you can also put out much more āAI slopā instead, which gets less clicks or views per item, put in total you get more clicks per effort put inā¦.
Nice clip. However, at 4:11, the ājust be good to me a capellaā already sounds dissonant to my ear to the underlying track you see, the underlying track is in Bb melodic minor, and starts with a Bb5 chords. On the same time, Lindy Layton starts singing +1 transposed in Ab minor. The Ab note isnāt in key Bb melodic minor, and thus sounds dissonant. It would have sounded better if the a capella were transposed to +3 overall.
The last mix from āShimmy Shimmy Yaā to āHigherā too, but as this is just a transition its doesnāt disturb (honestly, many of my mixes are the same), but persisting a dissonant combination like the Just Be Good To Me a capella distracts a lotā¦
Other than that, that guy has incredible technique
You got a fair point there: how many people in the crowd really hear this dissonance? If you make a bad mix, a lot of people will not notice it. Same goes for coverbands. While re-arranging stuff we often asked ourselves: will the audience notice? Probably notā¦ If I messed up a guitar solo, who cares? If I struck a note a few milliseconds earlier than the rest of the orchestra, who cares (apart from the director looking crossed at me)? Maybe my ear started to be too refined to just enjoy musicā¦ Maybe the shame is all on me for being too critical, and not just enjoying things
But that must not be an excuse to stop honing perfection. If perfection didnāt matter, I, as an amateur musician, might as well be playing guitar on rock festivals instead of Metallica (minus the drummer, some sayā¦). Or maybe I could just play bass in the national philharmonicā¦. Or You might just take the place of Swedish House Maffia on Tomorrowlandās line-upā¦ Or we might as well surrender to AI generated art formsā¦ The audience doesnāt hear the difference anyway, so why oppose to AI?
its the sum of a bunch of little mistakes like this that makes us amateurs, or the absence of these little mistakes, combined with a human touch that makes one a genius artistsā¦ If we stop caring for things like dissonance on a national radiostation, and if the audience stops caring, well, then AI slop has all the right to take over, because fine art isnāt appreciated anymore anywayā¦ Lets just pump out junk and monetise it
But the issue in that mix is only wrong in your personal opinion, if others think it sounds fine then has he done anything wrong? Its far from ājunkā in my humble opinion.
In your original comment you complain about people rigidly following the camelot wheel, but now ive demonstrated an example of someone breaking from that mould, thats a problem too.
Iām with Johan on the Just Be Good To Me mix. I can hear itās not right.
OTOH I disagree that the Camelot Wheel ādoesnāt workā. Itās not a new thing. Itās basically another name for the circle of fifths, which has been around since the 17th century.
Personally I find that mixes based on the system work well, even using the āfuzzyā method (ignoring A or B)ā¦most of the time. Generally though, Iām mixing intros/outros/breaks/bridges rather than full mix over full mix.
I think over the decades Iāve developed an ear for clashes, so if I hear something off, Iāll refrain from using a running mix.
That stuff that the Wu-Tang clan did with Texas though. Ouch!
I donāt want to bash this particular DJ: he is technically far more advanced than me. Hats of for this guy. But that one mistake is a typical example of what I hear on mainstream radio channelsā¦ everyone is trying to do STEM based mashups, but I learned today that many people donāt even look at the key column to throw a mashup togetherā¦ sticks out like a sore thumbā¦ and of course, the key column can be wrong, and there are exceptions to the circle of fifths: use your earsā¦
The bigger question is, how many people will hear there is something wrong: 0.1%? 1%? 5%? 10%? (We should have done a poll about this )
But lets be real, if really only 0.1% or even 1% of an average audience hears the difference between consonance and dissonance, and given that only 4% is really tone deaf, that would mean 95% is not really listeningā¦ If that 1% is the average appreciation of well written or played music, well, then its easy for some savvy IT guys to generate random AI junk, and monetise it. Iāll be hiding under a rock by the time this becomes mainstream.
Well, let me rephrase: in some cases, the circle of fifths does not work. All depends on the chord progression of the songs. If composer if piece A decided to go a different direction than the composer of piece B, they might be in the same key, but the chosen chords could still clashā¦ if I play a I, and you play a IV in the same key, this will not be consonantā¦
Also the circle of fifths it originally used to decide which borrowed chords or modulations would work. Which is a different application than throwing 2 songs over each other. Turns out most songs in the same key do fit over each other, because most EDM composers start with the I at bar one, and return to I at bar 5ā¦. And the iii and a V fit over a I, and the vi, IV and ii also fit togetherā¦
But I honestly was under the assumption that DJs producing mixes for the radio were not that stupid to complete ignore the key of their mashups, and I was hearing the exceptions to camelot that could happen depending on the chord progression. Turns out, as I harmonically analyzed @STU-Cās example, there are super technical DJs who have the audacity to completely ignore the circle of fifths.
I think the more likely scenario with the MNM mixes is that the DJ is NOT using the Camelot Wheel, or is maybe āgoing fuzzyā when it doesnāt work by ear.
Itās certainly the case that some people just cannot hear obvious key clashes. Iām aware of one Twitch DJ for example who will mix everything, regardless of how badly it clashes.
If however youāre mixing note over note, rather than chord over chord, as I said regarding where to mix, itās less of an issue.
Note over note will indeed always work, as long as youāre in the same keyā¦ its the basic principle of voice leading: if you have 2 melodyās sounding good on their own, in the same key, and join them, you automatically get chords that make sense.
Many of these mixes work because heās flipping over to the next beat and either echoing out or cutting out with the cross fader. Things donāt need to be in key when youāre mixing hip hop like that.
I also agree when he was blending into that The Game track it sounded off key but not terrible. A little dissonance is ok in a mix and sometimes thatās what youāre going for to create a moment.
If it was house music you wouldnāt mix it like that. There would be longer blends and if they were melodic parts they would likely need to be in key or close to sound good.
Truth is thereās many styles and there isnāt a one size fits all way to go about mixing.
I thought that Ghost-producing, Random Synthesizer Preset Generator (Not ai), midi file format conversion for notes, someone stems, samples for remixing, and using ONLY an analog mixers to making a actual sound loop (Check out āAnalog mixer is your synth?ā youtube videos) are enough to make music production already easier and faster in that workflow. But actual AI is too lazy.
The other problem is, people who also working at ghost producing losing clients and deserved money.
I was working yesterday listening to the Big Dance Energy playlist on Tidal and out of nowhere it just starts playing the most basic shitty techno with non-stop generic vocals. (Flying in the night sky like a butterfly) Immediately lost my flow and had to see what the heck was going on, apparently Tidal has a feature called Spotlight where their editorial staff are supposed to spotlight smaller artist but instead itās seemingly just an algorithm that promotes AI slop. This was the first time Iāve ever had to block and āartistā in the app.
you see, the underlying track is in Bb melodic minor, and starts with a Bb5 chords. On the same time, Lindy Layton starts singing +1 transposed in Ab minor. The Ab note isnāt in key Bb melodic minor, and thus sounds dissonant. It would have sounded better if the a capella were transposed to +3 overall.