Farid Zaripov
Two numbers describe where a track sits musically: its tempo in beats per minute, and its key. DJs need both to mix, producers need them to layer samples, and anyone learning an instrument needs them to play along. Both can be measured by ear, both can be detected automatically, and automatic detection fails in specific, predictable ways worth understanding.
Tempo: Beats Per Minute
BPM counts the pulse you tap your foot to. Rough genre ranges:
| Genre | Typical BPM |
|---|---|
| Hip-hop | 80–100 |
| Pop | 100–130 |
| House | 120–130 |
| Techno | 125–150 |
| Drum and bass | 165–180 |
| Ballads | 60–80 |
Counting It by Hand
Tap along with the beat for 15 seconds and multiply the count by four. Tapping for 30 seconds and doubling is more accurate; a full minute is more accurate still but harder to stay locked to. The result will land within a beat or two of the true value, which is enough for practice, though not for beatmatching.
The Halving and Doubling Problem
The most common disagreement between people, and between software, is a factor of two. A track at 140 BPM with a strong backbeat is frequently heard — and reported — as 70. Neither is wrong: they are the same pulse counted at different metrical levels.
The convention that settles it is genre. Drum and bass is written at 174, not 87, even though the bassline often moves at half that. House is 128, not 64. When automatic detection returns something that feels wrong, check whether doubling or halving it lands in the expected range for the style — that resolves the majority of cases.
Where Automatic Detection Struggles
- Rubato and live performance. Music that breathes has no single BPM. Classical recordings and live takes can drift several BPM within one phrase.
- Sparse intros. Detection needs transients; a pad-only opening gives it nothing to lock onto.
- Half-time and double-time sections. A track that switches feel mid-way produces one value for the first half and another for the second.
- Swing. Heavily swung rhythms confuse algorithms tuned for even subdivisions.
- Complex time signatures. Anything not in 4/4 tends to yield an odd multiple.
For modern produced music built to a grid, automatic detection is essentially always right. A BPM and key finder analyzes the whole file rather than a fifteen-second sample, which also catches tracks that change tempo partway through.
Key: Which Notes the Track Lives In
Key describes the tonal center — the note everything resolves toward — and the mode, most often major or minor. Detection works by measuring how much energy each of the twelve pitch classes carries across the track, then matching that profile against known key templates.
Why Detection Is Harder Than Tempo
- Relative major and minor share every note. C major and A minor contain identical pitches, differing only in where the music resolves. Confusing the two is the single most common error, and it matters — mixing C major into A minor is fine, but mislabelling changes how you read the rest of the wheel.
- Modulation. Tracks that change key mid-way get reported as whichever section dominates.
- Modal music. Dorian, Mixolydian and other modes get flattened into the nearest major or minor.
- Percussive and atonal material. Drum loops and noise-based tracks have no key. A detector will still return one, because it always returns its best match — and that answer is meaningless.
- Heavy processing. Distortion and saturation add harmonics that skew the pitch profile.
Treat a detected key as a strong hypothesis rather than a fact. The confirmation takes seconds: play the reported tonic note against the track and listen for whether it sits at rest or fights.
The Camelot Wheel
DJs rarely say "F sharp minor". They say 11A. The Camelot system maps all 24 keys onto a clock face where every position is a fifth from its neighbors, and the letter marks the mode: A for minor, B for major.
| Camelot | Minor (A) | Major (B) |
|---|---|---|
| 1 | A♭ minor | B major |
| 2 | E♭ minor | F♯ major |
| 3 | B♭ minor | D♭ major |
| 4 | F minor | A♭ major |
| 5 | C minor | E♭ major |
| 6 | G minor | B♭ major |
| 7 | D minor | F major |
| 8 | A minor | C major |
| 9 | E minor | G major |
| 10 | B minor | D major |
| 11 | F♯ minor | A major |
| 12 | D♭ minor | A♭ major |
The point of the arrangement is that compatible moves become arithmetic:
- Same number, switch letter (8A → 8B) — relative major/minor. Same notes, different emotional color.
- ±1, same letter (8A → 9A or 7A) — a fifth away. The standard smooth transition.
- ±2, same letter — noticeable but usable, often as a deliberate lift.
- +7 semitones or a jump across the wheel — a clash unless it is intentional.
Energy mixing extends this: moving up a number raises tension, and a +1 move combined with a switch from minor to major is the classic way to build a set.
Changing One Without the Other
Historically, speeding up a record raised its pitch — the two were physically coupled. Digital tools separate them, but neither operation is free:
- Time stretching changes tempo at constant pitch. Beyond roughly ±10 % it introduces a smeared, watery character on sustained sounds and flams on transients.
- Pitch shifting changes key at constant tempo. Beyond about ±3 semitones, voices start to sound synthetic — the formants shift along with the pitch, which the ear reads as a different-sized singer.
For DJ use, small tempo adjustments of ±6 % are inaudible on most material. For production, the honest rule is that a sample shifted more than a few semitones announces itself.
A Practical Workflow
- Detect automatically to get both numbers in one pass.
- Sanity-check the BPM against the genre range; double or halve if it looks off by a factor of two.
- Verify the key by playing the tonic against the track, and specifically test the relative major or minor if it feels ambiguous.
- Convert to Camelot if the purpose is mixing.
- Note anything that changes mid-track — a tempo shift or modulation matters more than the headline value.
For transcription rather than mixing, the next step is turning the audio into notes: note detection extracts the pitches themselves, which is what you need to play a part back rather than merely align two tracks.
