How to find a song's BPM, and why you keep getting two answers

Tap along to a track four times and you will get four different numbers. That is not a failure of your ear, it is a real ambiguity: a song at 85 BPM and the same song at 170 BPM are mathematically the same music counted differently, and both answers are defensible.

Knowing which one to write down is most of the skill. Here is how to find a tempo three different ways, when each is right, and what to do with the number once you have it.

Method one: tap it, and handle the halving problem

Any tap-tempo tool works: tap the beat, it averages your intervals, it shows a number.

Two rules make it accurate.

Tap at least sixteen beats. Four taps carry your reaction time as a large error. Sixteen averages it away.

Decide what you are tapping. Tap where you would nod your head, not every hi-hat. The half-time and double-time confusion comes entirely from tapping a subdivision instead of the pulse.

Then sanity-check the result against the genre. If you tapped a hip-hop track and got 170, you counted double: that is 85. If you tapped drum and bass and got 87, you counted half: it is 174. Genres cluster hard around known tempos, and that cluster is your error correction.

Method two: count against a clock

No tools, works anywhere. Count beats for fifteen seconds and multiply by four.

Slightly less precise, entirely reliable, and it forces you to commit to what the beat is, which is exactly where tap tools let you fool yourself.

Method three: let software detect it

DAWs, separation tools like Moises and dedicated analysers all do automatic tempo detection, and they are good on material with a steady grid.

Where they fail is predictable: live-recorded music that drifts, rubato passages, tracks with a tempo change, and anything where the strongest transient is not on the beat. Automatic detection also makes exactly the same half-time and double-time error you do, so the genre sanity-check still applies.

Typical tempo ranges, and why they matter

Two practical uses for this table.

Error correction. A number outside its genre's range by roughly double or half is a counting error, not a discovery.

Prompting. Asking a generator for a genre and a contradictory tempo produces a fight the drums usually lose. Ask for thrash at 100 BPM and you get something confused rather than something slow and heavy.

What to do with the number

In a generated brief, do not lead with it, and do not rely on it alone. This is the part that surprises people: Suno does not treat BPM as a hard parameter. A bare number is a weak signal it can round off or ignore, because the model was trained on descriptions of music rather than on a metronome. Our own guide to controlling tempo goes through the evidence.

What works is stacking signals. Pick a genre whose native tempo already is your target, name the groove so the pulse is described rather than labelled, and put the figure at the end as a confirmation rather than a command. Every recipe in our catalogue does exactly this: Thrash metal, 1980s, down-picked distorted guitars, galloping bass, double-kick drums, 170 BPM pins the speed four separate ways before the number appears.

Move it deliberately, not accidentally. The same recipe at 96 and at 118 reads as the same identity with different energy, which is how you build a series of tracks that belong together. Changing instruments instead changes the identity.

Match it when layering. If you are putting generated music under an existing edit or over a loop you already have, the tempos need to agree before anything else can.

Do not treat it as the whole feel. Two tracks at 120 BPM can feel completely different depending on subdivision and swing. Tempo sets speed; the drum pattern sets the feel.

Half-time, double-time and why producers argue about it

The most common genuine disagreement in tempo is not an error at all.

A track written at 140 BPM with a snare on beat three rather than beats two and four feels like 70. Trap sits at 140 with hi-hats subdividing so fast the pulse reads as 70. Both numbers describe the same music, and which one you write down depends on what you are doing with it.

Write the higher number when you are programming, because your grid and hi-hat divisions will make sense.

Write the lower number when you are describing feel to a person or a generator, because that is what the body responds to.

When a generated track comes back feeling wrong at the tempo you asked for, this is usually why: you asked for the number and got the other reading of it. Specifying the drum pattern alongside the tempo, half-time snare, driving four-on-the-floor kick, resolves it in one token.

Tempo in a generated track that drifts

Generated music sometimes wanders off the grid, which shows up when you try to sync it to a video or loop it.

Three things help:

Ask for a steadier feel explicitly. Terms like tight programmed drums or machine-locked groove push toward stability; loose live feel pushes away from it.

Keep sections short. Drift accumulates. Controlling structure with section tags limits how far it can wander before a new section resets the feel.

Fix it in an editor rather than regenerating. Time-stretching a good take beats rolling for a perfectly timed one, and regeneration costs credits that do not roll over.

FAQ

What is the fastest way to find a song's BPM? Tap along for sixteen beats with any tap-tempo tool, then check the answer against the genre's usual range.

Why do BPM finders give different numbers? They disagree about what counts as the beat. A doubled or halved result is the same tempo counted at a different subdivision.

Is 128 BPM fast? It is standard dance tempo. Fast for a ballad, slow for drum and bass.

Can I change a song's BPM without ruining it? Modest time-stretching, within roughly five percent, is usually inaudible. Large changes distort transients and vocal formants.

Does BPM matter in an AI prompt? Less than people assume. Suno treats a bare number as a weak signal. Feel words and a named groove pin the tempo harder, with the figure added at the end as confirmation.

How do I find BPM without any tool? Count beats for fifteen seconds, multiply by four, then check against the genre range.