Why Does My Library Say Drum & Bass Is 87 BPM?

You import a drum & bass collection, look at the tempo column, and every track reads somewhere around 87 BPM. Not 174. Not close to 174. Half of it, on tracks you have been mixing at 174 for years.

Nothing is broken, in your library or in the files. What you are looking at is the single most common failure in automatic tempo detection, and it is worth two minutes of your time because it changes how you search, sort and filter every track you own.

The short answer

The analyser locked onto the half-time pulse instead of the written tempo. 87 × 2 = 174. Your tracks are 174 BPM tracks, and the number in the column is a real measurement of a real pulse in the music — just not the one the genre is written in.

This is called a tempo octave error, and "error" oversells it. Doubling or halving a tempo produces a grid that still lines up perfectly with the music, which is why a half-time reading beatmatches fine and why nobody notices until they sort a crate by BPM.

Why detectors do this, specifically in drum & bass

A tempo detector doesn't read a producer's project file. It looks for periodic structure in the audio and picks the strongest repeating interval it can find. In most dance music that is unambiguous — a four-to-the-floor house track puts a kick on every beat, so the loudest, most regular thing in the file is the tempo.

Drum & bass is built the other way round. At 174 BPM the pattern is a kick on beat one and a snare on beat three, which means the loudest low-frequency events in the track alternate at exactly half the written tempo. The 174 lives in the hi-hats and the chopped break — quieter, denser, harder to lock onto. So a detector weighing the biggest transients finds a rock-solid 87 BPM pulse and returns it with high confidence, because it is a rock-solid 87 BPM pulse.

That same relationship is why the half-time trick works as a mixing technique at all: a hip-hop track at 87 and a drum & bass track at 174 share a downbeat, which is the whole basis of the half-time / double-time bridge. The feel is deliberate and musical. The label is the accident.

How common is it? We counted

Across our public catalog of 443,201 analysed tracks, measured 10 August 2026, 46,678 tracks (10.5%) are stored below 95 BPM — and most of them are not slow music. They are fast music read at half speed.

81.9% of drum & bass is affected. Not a fringe of unusual files: four tracks in five. Compare tech house at 0.9%, which is what a genre looks like when the kick and the tempo agree.

The histogram is the clearest way to see it. There should be one pile around 174. There are two piles, the enormous one is in the wrong place, and there is almost nothing in between — 17,944 tracks in the 85–90 bucket against 725 from 165 BPM upwards. That gap matters: a distribution with two separated modes cannot be repaired by throwing away outliers, because the outlier test keeps whichever mode is bigger. Here the bigger mode is the wrong one.

One more measured detail worth knowing, because it is counter-intuitive. Catalog-wide, the damage is entirely in the tails and in the per-genre view: the median tempo across the whole catalog is 125 BPM raw, and 126 BPM if you throw out everything under 95. Removing a tenth of the catalog moves the headline number by one BPM. It is genre pages, tempo searches and BPM-sorted crates that get wrecked, not averages.

How to check your own library in about a minute

1. Sort by BPM, ascending. Everything genuinely slow floats to the top — intros, ambient, downtempo, a cappellas.
2. Look for a cluster you know is fast. A block of drum & bass at 85–90, jungle sitting in the 80s, or dubstep at 70 rather than 140. A cluster is the tell; one odd track is usually just an odd track.
3. Double it and sanity-check the genre. 87 → 174 is drum & bass. 70 → 140 is dubstep. 105 → 210 is nothing, so leave that one alone — this is exactly the test that stops you "fixing" a hip-hop track at 87 that really is at 87.
4. Check the grid before you re-analyse. A half-time reading usually has a correct beatgrid, just labelled at half density, so the track will still mix. If you want the number fixed, DJ software exposes both an analysis BPM range in its preferences and a per-track ×2 / ÷2 toggle — the toggle is instant and doesn't destroy your cue points.

Whether it is worth fixing depends on how you work. If you find tracks by browsing playlists it changes nothing. If you filter by tempo range, use BPM-based smart crates, or hand your library to anything that reads the BPM field, a fifth of your bass music is invisible to your own filters.

What we did about it, and when

Until 10 August 2026 this bug was live on our own site. Our drum & bass tempo page published "drum & bass sits at 87 BPM at the median" against a stated editorial range of 170–180, and described that as a tight distribution. It was the same failure this article describes, published as a fact. We fixed it that day and this is the rule we shipped, stated in full so you can disagree with it:

• A genre gate. The correction only runs on a genre where more than 5% of tracks sit under 95 BPM. A handful of strays in a clean genre is not a half-time problem.
• A per-track arithmetic test. A track is doubled only if it is under 95 BPM and twice its stored value lands inside that genre's editorial range, widened by 15%.
• Applied before the outlier filter, never after. Doubling first collapses the two piles into one, so the outlier filter then does the job it was written for instead of picking a side.

Both halves of that are load-bearing. A naive "double everything under 95" moves hip-hop's median from 105 to 134 and reggaeton's from 96 to 119 — it would trade one wrong number for two. Under the shipped rule those genres take zero corrections, because doubling a 70 BPM hip-hop track lands at 140, nowhere near hip-hop's 80–100 range. That is what separates a correction from a fudge: it has to decline to fire.

Nothing was written to the stored tempo of a single track. The correction lives in the display layer, so what you see on a genre page is corrected while the recommendation engine still reasons about the raw analysis — deliberately, because our catalog stores drum & bass the same way Beatport, Rekordbox, Traktor and Serato display it, and quietly rewriting 18,000 rows would break every tempo query in the product. Fixing the analyser itself is a separate job and it is not done.

How to read our tempo pages now

• Every genre tempo page states its corrections. Under the answer on /bpm-for/drum-and-bass is a line giving the date, the number of tracks measured, how many outliers were dropped and how many were detected at half tempo and doubled — 17,776 of them. If a page says no track needed a half-time correction, that is a measurement, not a silence.
• Tempo listing pages still show what is stored, and say so. /bpm/87 holds 10,946 tracks, 9,964 of them tagged drum & bass, while /bpm/174 holds 173 — a 63:1 inversion of reality. Aggregating that away would be a lie about what the catalog contains, so the page declares itself instead: most of these are 174 BPM tracks detected at half tempo.
• Don't trust a "needs review" flag to find these. Our detector flags 61.1% of sub-95 tracks as uncertain against 10.8% of everything else, which sounds like a usable signal — but it fails on the exact case you care about. Drum & bass corrects 17,776 tracks and flags only 4,405, because the validation range for the genre accepts both 85 and 174 as plausible. The flag catches tracks that look strange. A half-time drum & bass track looks completely normal.

If you want the arithmetic for bridging tempos on purpose rather than diagnosing them, the half-time / double-time converter does it, and the mashup trick covers using the relationship musically. For where drum & bass actually sits and what it mixes with, the drum & bass mixing guide and the BPM guide are the ones to read.