Free · no login · nothing uploaded
DJ ear training.
Two drills for the two things a DJ hears before they touch anything: how fast is this, and what is it against the record already playing. Scored in the units printed on the equipment — percent of pitch, and notes in common.
Tempo
A loop, at a tempo it will not tell you.
Listen, then commit to a number. You will be scored on how far the fader would have had to move, not on how many you got right.
Where you are
Nothing yet on this level. After a handful of goes this becomes your median miss in percent of pitch, which is the number that says whether you could set a fader from what you heard.
Level
House through techno, whole numbers. The range most sets live in.
Listening
History
Kept in this browser and nowhere else. Clearing it starts the diagnosis again.
The wheel
Every rule falls out of one number.
It is usually taught as a picture to memorise. It is not a picture, it is a count.
The Camelot wheel looks like a convention — these keys mix, those do not, learn the diagram. It is not a convention. It is a count of how many of the seven notes two keys have in common, and every rule anybody has ever given you about it is that count in disguise.
| Move | From 8A | Notes in common | What it does |
|---|---|---|---|
| Same key | 8A → 8A | 7 of 7 | Nothing to clash. Everything lines up, and long blends stay clean the whole way through. |
| Relative major / minor | 8A → 8B | 7 of 7 | Nothing added and nothing taken away — only the weight moved. It is the lift you feel when a track opens up without changing key. |
| Up a fifth | 8A → 9A | 6 of 7 | One step clockwise. Brighter by exactly one note, which is why it reads as energy going up. |
| Down a fifth | 8A → 7A | 6 of 7 | One step anticlockwise. Same size of move, opposite direction — the room settles instead of lifting. |
| Two round the wheel | 8A → 10A | 5 of 7 | It works, and it is a jump rather than a step — you would use it to change the room, not to hold it. |
| Same root, other mode | 8A → 11B | 4 of 7 | Same home note, opposite third. The bass agrees and the chords argue, which is a very particular kind of wrong. |
| A semitone out | 8A → 3A | 2 of 7 | Nothing else in mixing sounds like this, and nothing else is this unmistakable. |
The two sevens at the top are the interesting pair, and they are the reason this is a drill rather than a lookup. A minor and C major are the same seven notes — nothing was added and nothing taken away — and they still sound like different places to stand, because what moved was which note the music treats as home. No count can hear that for you.
The clash
One record in 12 is a semitone out
Not a rare accident. A structural feature of the tempo you happen to play at.
Take any tempo. Take two records at it, at random, from the 443,544 we hold with a key on them. 8.3% of the time they are a semitone apart — one pair in 12 — and only 18.2%, about one in 5.5, are a move the wheel would sanction.
The interesting half is what the catalog's own lopsidedness does to that, which is nothing. Electronic music is not written evenly across the twenty-four keys — 61.2% of it is minor, A minor alone holds 8.13% of everything and A major holds 1.65% — and that skew is real enough to make two records land in the same key 17.1% more often than chance. It leaves the semitone clash at 8.31% against a chance figure of 8.33%.
So playing in the popular keys concentrates your good luck and does not touch your bad luck. The clash arrives at exactly the rate it would in a world where everybody wrote in every key, which is the argument for being able to hear it rather than trusting where your crate happens to sit. Measured over 41 whole-BPM buckets between 118 and 178, each holding at least five hundred records, so it is not one tempo's quirk.
The unit
Why the tempo score is a percentage.
Four BPM is not four BPM. It depends entirely on what it is four BPM of.
Every BPM guessing game marks you in BPM, and BPM is the wrong unit for the same reason a percentage was the wrong unit on the beatmatch trainer: it does not mean the same thing twice. Being four out at 174 is a smaller move of the fader, a smaller change to the record, and a smaller mistake than being four out at 122 — and a drum and bass DJ is not worse at this than a house DJ because the numbers they work with are bigger.
| Out by | At 122 | At 128 | At 140 | At 174 |
|---|---|---|---|---|
| 1 BPM | 0.82% | 0.78% | 0.71% | 0.57% |
| 2 BPM | 1.64% | 1.56% | 1.43% | 1.15% |
| 4 BPM | 3.28% | 3.13% | 2.86% | 2.30% |
| 8 BPM | 6.56% | 6.25% | 5.71% | 4.60% |
Under about 1.5% nobody in the room hears anything. By 3% a vocal or a recognisable riff has audibly moved. By 6% you are outside the range most players even offer without a wide switch, and the record has changed character. Those are the bands the verdicts use, and they come from what a pitch move does rather than from round numbers.
Half time
87 and 174 are the same record
The one answer this refuses to mark wrong, and the catalog is why.
Count the kick in a drum and bass record and you get one number. Count the snare and you get half of it. Both are honest, and the disagreement is not a detail — the median published tempo of the 23,185 drum and bass records we hold is 87 BPM, for music that is played, without exception, at around 174. That is what the labels and distributors filed.
Our own analyser, listening to the audio rather than reading the metadata, disagrees by exactly a factor of two on 12,787 public tracks in one direction and 5,538 in the other, out of 443,212 compared. Neither party is wrong on any of them.
A trainer that marked 87 wrong on a 174 breakbeat would therefore be teaching a DJ that the industry's own metadata is a mistake. So the octave reading is detected, named, scored against the tempo your ear actually produced, and counted on its own line. What matters in a booth is not which of the two you say — it is knowing that both exist, so a track filed at 87 does not get skipped by somebody building a set at 174.
The evidence
What was measured, and how.
A key trainer that is out of tune marks right answers wrong, and you would have no way to tell.
The chords here are not samples. They are three notes worked out from one expression — 440 × 2(m−69)/12 — which means the whole question of whether the tool is in tune is answerable rather than assertable. So it was answered: every one of the 24 keys was rendered offline from the exact file your browser loads, and the frequency of all 72 notes read back out of the resulting audio.
Worst deviation from equal temperament: 0.575 cents, averaging 0.444, at 48.0 kHz. A cent is about the smallest pitch difference a trained listener can pick out under ideal conditions and roughly a twentieth of what anybody notices in music, so the worst chord in the tool is out by half of the smallest thing anybody can hear. Every minor third measures three semitones and every major third four, to within 0.0023 of one.
The reading is a spectral centroid rather than a peak, and that distinction is not pedantry — a peak-finder reported up to 12.7 cents on the same audio, always on the higher notes, growing with frequency and then flipping sign, which looks exactly like a transposition bug and is not one. A detuned pair is two lines rather than one, and over a short decaying window the peak lands on whichever of them the window favoured. The estimator is checked against a deliberately planted twenty-cent error, because a centroid taken around the frequency you asked about will happily report zero for silence.
The same rig proves the loops themselves: no chord is ever cut short by another voice, and two pitched decks a semitone apart peak under full scale — worth checking, because clipping on the loudest beat does not sound like clipping, it sounds like a slightly nastier chord.
How to practise
Anchor, then interpolate
Both of these get much faster with one specific technique, and it is not listening harder.
Tempo. Nobody hears 128 in the abstract. What people have is one tempo they know cold — the one they play most nights — and everything else is measured against it. So do not try to name the number. Ask whether it is faster or slower than your anchor and by roughly how much, and convert. If you have no anchor, build one: put the metronome on at the tempo you play at, listen for a minute, and come back. The bias figure in the panel is the fastest thing to fix — guessing consistently high is one correction, not a skill.
Key. Start on the warm-up level, which asks one question: is this in key or a semitone out. Two notes in common against seven is the widest gap on the wheel and if that one goes past you, nothing smaller will land. Use the cue button — pull the second deck out and put it back rather than listening harder to both at once, because a clash announces itself on the way in far more clearly than it does once you have got used to it. Only move up when the clash is automatic.
The relationship most people find hardest is not the clash, it is telling a fifth from a relative major, and the reason is in the table above: they share six and seven of the seven notes. The panel will tell you if that is you, which is the entire reason it is a table and not a score.
Honestly
What this cannot teach you.
What it is not, so nobody finds out the hard way.
It plays loops, not records. A loop has no vocal, no breakdown and no key change halfway through, and all three of those are places a real harmonic judgement gets hard. It also plays one chord shape — a triad — where a record has a bassline, a pad, a vocal and a sample that may disagree with each other about what key it is in. If you want the difficult version, put two of your own records on and use the key compatibility checker afterwards to see whether you were right.
The tempo drill cannot teach you to beatmatch either, and it is worth being clear about why: knowing a record is 128 does not get you a mix, because being one whole BPM out is already a wrecked blend inside two bars. Hearing the tempo is how you decide whether a record belongs in the next twenty minutes. Matching it is the beatmatch trainer, and it is a different exercise with a different unit.
Nothing you do here leaves your machine, nothing is uploaded, and there is no account. Your attempts are kept in this browser and nowhere else.
Frequently asked
Common questions
Can you learn to hear BPM?+
Yes, and much faster than people expect, because it is not absolute pitch — it is calibration against tempos you already know cold. Almost every DJ can already recognise the tempo they play at most nights to within a couple of BPM; the skill is spreading that anchor outwards. Twenty attempts a day for a week moves most people from guessing to being reliably inside a couple of percent, which is the point at which you can hear a record in a club and know whether it belongs in your next hour.
Why is the score in percent instead of BPM?+
Because BPM does not mean the same thing twice. Four BPM out at 128 is 3.13% of pitch and four BPM out at 174 is 2.30% — the second is a much smaller move of the fader and a much smaller mistake, and a score in BPM says they are identical. Percent of pitch is also the unit already printed on every deck ever made, so the number lands somewhere you have an instinct for: under about 1.5% nobody in the room hears anything, by 3% a vocal has moved, by 6% it is a different record.
I said 87 and it was 174. Is that wrong?+
No, and the tool says so. That is the same record counted at half speed, and it is the reading the industry itself keeps — the median published tempo for the 23,185 drum and bass records in our catalog is 87 BPM, for music every DJ on earth plays at 174. Across the whole catalog our own analyser reads exactly double the published figure on 12,787 tracks and exactly half on 5,538. Half-time answers are detected, named and counted on their own line — never averaged into your error and never marked wrong.
Do I need perfect pitch for the key drill?+
No, and it would barely help. The drill never asks you to name a key — it asks what the second record is against the first, which is a relative judgement and the only one that matters in a booth. Nobody stands over a deck thinking "that is F minor". They think "that fits" or "that does not", and the whole of harmonic mixing is that one judgement made quickly.
What is a semitone clash?+
Two records a semitone apart. It is the single most audible harmonic failure in a mix, because a semitone leaves only two of the seven notes in common — every other relationship on the wheel leaves at least four. It is also not rare: at any given tempo in our catalog, one record in 12 is a semitone from the one playing, which is almost exactly what chance would give. The key skew in electronic music does not protect you from it.
Why does the tool not show a streak?+
Because a streak cannot tell you what to practise. Somebody scoring 70% is not seven-tenths of a DJ — they are somebody who hears five of the relationships cleanly and swaps the other two, and until they know which two the practice has nowhere to go. So the key drill keeps a table of what you said against what it was, and the tempo drill keeps a median miss with a bias, because guessing consistently high is a fixable habit and guessing randomly is not.
How do I know the tool itself is in tune?+
Because it was measured rather than claimed. Every chord it can play was rendered offline from the exact file your browser loads, and the frequency of every note in it read back out of the audio: 72 notes across 24 keys, all of them within 0.575 cents of equal temperament at A = 440, averaging 0.444. A cent is roughly the smallest pitch difference a trained listener can detect under ideal conditions, so the worst note here is about half of one. Any tuning you hear is the music, not the tool.
Can I use this with headphones on one ear?+
Yes — turn on split cue and the two decks go to opposite ears, which is how you would actually hear them in a booth. It makes the key drill noticeably easier at first because the two chords stop fighting for the same space, and then noticeably harder to leave behind, so it is worth doing some attempts without it.
Why can I not tap the tempo out?+
Because tapping is measuring, not hearing, and the two are different skills with different uses. When you genuinely want to measure a tempo there is a tapper for it and it is better at the job than your first impression would be. The reason to be able to hear it instead is speed: a tap needs eight bars and a decision needs one.
Does it keep my results?+
Yes, in this browser and nowhere else. The diagnosis is built from every attempt you have made on the level you are on, so it gets more useful the longer you leave it alone. Nothing is uploaded, there is no account, and clearing your browser data clears it.
Explore
Beatmatch trainer · DJ metronome · Camelot wheel · Harmonic mixing chart · Key compatibility · BPM tapper · Camelot key · Relative major & minor · Harmonic mixing · All tools
See which tracks actually fit this one
DJ terms on this page: Camelot key · Harmonic mixing · Relative major and minor · Musical key · BPM · Pitch fader · Open Key · Key lock · Full glossary