Why does the same song show different BPMs, and which reading is right?
Short answer: one beat can be counted at more than one speed, so a BPM finder has to pick one, and different excerpts of a song can give slightly different numbers. We made 35 test songs at tempos we set ourselves and dropped them into this site's BPM finder:
- The main number was the built-in tempo on 18 of 35.
- On 14, the right tempo was the second reading under it, marked half time or double time (our 70 BPM songs read 140, our 160 and 174 BPM songs mostly read 80 and 87).
- On 3, the main number was 2/3 or 4/3 of the tempo (174 read as 116.0), and the right tempo was in the "Other tempos that also fit" line.
So the right tempo was somewhere on the page every time, but you choose which. Press Play with clicks, switch between the two readings, and keep the one where the clicks land on the beat at the speed you count. Use at least 10 seconds of the part of the song with drums; an intro without drums read half the tempo in our test.
Why one song has more than one BPM
A BPM finder does not hear beats the way you do. It looks for the moments when sound starts (a drum hit, a plucked note) and then for a time gap that keeps repeating. A song with a hit every half beat repeats at the half beat, the beat, two beats and a bar, all at once. Any of those gaps is a correct pattern; only one of them is the speed people tap their foot to, and that depends on how they count.
That is why 70 and 140 BPM, or 87 and 174, are the same grid of hits: one counts every hit on the hi-hat as a beat, the other every second one. A finder has to choose. The one on this site scores every tempo from 40 to 240 BPM and, among readings that fit equally, leans toward numbers near 120 (that preference is written in its code). Then it shows the other one as "half time" or "double time" so you can switch. A finder with a different preference would print the other number for the same file; neither is broken.
The 2/3 case is a different trap. With a kick on every beat and a hi-hat on every half beat, the pattern also repeats every three half-beats, one and a half beats apart. At 174 BPM that is 116 BPM, and it is not a half or a double, so switching to the second reading does not fix it.
How we tested
- Tool: this site's BPM finder, the real page run on a local copy in Chromium 153 through Playwright 1.63 (headless). Our script picked each file in the page's file box and copied the main BPM, the half or double reading, the "Other tempos that also fit" line, the confidence line and the tempo-change note.
- 35 test songs we made (44.1 kHz, 16-bit mono WAV, 30 s each): five drum patterns at 70, 85, 100, 120, 140, 160 and 174 BPM, each with sustained chords. The patterns: kick on every beat with claps on 2 and 4; a rock beat (kick on 1 and the "and" of 3, snare on 2 and 4); a half-time beat (snare only on 3, hi-hat every quarter beat); kick on every beat with a hi-hat every half beat and no snare; and a two-step breakbeat. Every pattern had hi-hats at least every half beat.
- One 3-minute song at 128 BPM with sections: a 30 s intro of chords only, 60 s of full beat, a 30 s half-time breakdown, then 60 s of full beat. We tested parts of it and excerpts of 5 to 60 seconds.
- One 3-minute song that speeds up evenly from 118 to 126 BPM, like a band rushing.
- One real recording: Mozart's Symphony No. 40 in G minor, first movement, played by the Musopen Symphony, in the public domain on Wikimedia Commons, converted to 44.1 kHz, 16-bit: the whole movement (5:58.8) and eleven 30 s excerpts. We have no true tempo for it, so for this one we only compare the readings with each other.
What the finder read on songs with a known tempo
| Built at | Kick every beat, claps | Rock beat | Half-time beat | Kick every beat, no snare | Breakbeat |
|---|---|---|---|---|---|
| 70 BPM | 140.0 | 140.0 | 140.0 | 140.0 | 140.0 |
| 85 BPM | 85.0 | 85.0 | 113.3 | 85.0 | 85.0 |
| 100 BPM | 100.0 | 100.0 | 100.0 | 100.0 | 100.0 |
| 120 BPM | 120.0 | 120.0 | 120.0 | 120.0 | 120.0 |
| 140 BPM | 70.0 | 140.0 | 140.0 | 140.0 | 140.0 |
| 160 BPM | 80.0 | 80.0 | 80.0 | 106.7 | 80.0 |
| 174 BPM | 87.0 | 87.0 | 87.0 | 116.0 | 87.0 |
Where the main number is right, it was right to the decimal shown (85.0, 100.0, 120.0). Every wrong main number was a simple fraction of the right one:
- Double (5 files): all five 70 BPM songs read 140.0, with 70.0 as the half-time reading.
- Half (9 files): one 140 BPM song, four 160 BPM songs and four 174 BPM songs, with the right tempo as the double-time reading.
- 2/3 (2 files): kick on every beat with no snare, at 160 (read 106.7) and 174 (read 116.0). The second reading (213.3 and 232.0) is no help here; 160.0 and 174.0 were listed in "Other tempos that also fit".
- 4/3 (1 file): the half-time beat at 85 read 113.3, with 85.0 first in "Other tempos that also fit".
The confidence line did not warn about any of these: all 35 readings said Strong beat, right or wrong. It tells you that a steady repeating pattern was found, not that the number is the one you count. The pattern also changed the answer at the same tempo: at 140 BPM, the song with a kick on every beat read 70.0 while the other four read 140.0.
Which part of the song you measure
| Excerpt | Built at | Main reading | Confidence |
|---|---|---|---|
| Intro, chords only (0:00–0:30) | 128 | 64.0 | Moderate |
| Intro into the beat (0:25–0:55) | 128 | 128.0 | Strong beat |
| Full beat (0:30–1:00) | 128 | 128.0 | Strong beat |
| Half-time breakdown (1:30–2:00) | 128 | 128.0 | Strong beat |
| Whole song (3:00) | 128 | 128.0 | Strong beat |
| 5 s of full beat | 128 | no reading: the page asks for at least 7 s | — |
| 7 s of full beat | 128 | 128.1 | Strong beat |
| 10, 15, 20, 30 and 60 s of full beat | 128 | 128.0 each | Strong beat |
| Speeding-up song, 0:00–0:30 | 118.0 → 119.3 | 118.7 | Strong beat |
| Speeding-up song, 1:15–1:45 | 121.3 → 122.7 | 122.0 | Strong beat |
| Speeding-up song, 2:30–3:00 | 124.7 → 126.0 | 125.3 | Strong beat |
| Speeding-up song, whole (3:00) | 118.0 → 126.0 | 124.0, with the note "Parts of this file read between 118.4 and 125.5 BPM" | Moderate |
On a song with a steady tempo, the section did not matter once the drums were in, even the half-time breakdown, and 7 seconds was enough. The intro without drums is the exception: slow chord changes are the only thing that repeats, and it read half the tempo with a weaker confidence.
On a song whose tempo moves, every excerpt gives a different, correct number for that stretch, and the whole file gives one number that is not the average (124.0 for a song that went from 118 to 126). The note about parts reading between two numbers is the sign to trust a range rather than one figure.
A real orchestra recording
The whole Mozart movement read 101.9 BPM (double-time reading 204.0, confidence Moderate), with the note that parts read between 97.7 and 104.2. The eleven 30 s excerpts read from 98.9 to 103.5, a spread of 4.7%: 8 said Moderate, 2 said Strong beat and 1 said Weak. The 6-minute file took 5.4 s.
No excerpt jumped to half or double; they just wandered by a few BPM, as a played tempo does. We cannot say whether 102 or 204 is the right count here, because the recording has no click to check against. For music played by people rather than a machine, expect a tempo per section and a few BPM of difference between excerpts, not one exact number.
How to get the reading you want
- Use the part with drums. Skip a quiet intro; cut 10 to 30 seconds of the main beat with the audio cutter if the song starts slowly, or drop in the whole song.
- Play with clicks on both readings. The finder shows the main number and the half or double one. Press Play with clicks, then switch. Keep the reading whose click matches the speed you tap your foot to; both will sit on hits.
- If neither fits, read the "Other tempos" line. When the clicks drift against the music on both readings, try the numbers listed there: in our test the right tempo was the first or second of them on all 3 files where the main reading was 2/3 or 4/3.
- Cross-check by tapping. Tap tempo on the same page works out the tempo from your first tap to your last; tap for longer to even out small timing errors. If your taps land near one of the numbers on the page, use that one.
- Look for the tempo-change note. If it appears, the song does not have one BPM. Measure the section you need on its own.
Questions
Why do two BPM finders give different numbers for the same song?
If one number is about half or double the other, both describe the same beats, counted differently, and each finder has its own preference for which to print first. A difference of a few BPM can come from measuring different parts of a song whose tempo moves. In our test the finder on this site was exact to 0.1 BPM on steady machine-made tempos, and wandered by up to 4.7% between excerpts of a live orchestra.
Is "Strong beat" a sign the number is right?
No. It means the pattern repeats clearly. In our test all 35 songs with a known tempo said Strong beat, including the 17 where the main number was not the built-in tempo.
How long a clip do I need?
The page needs at least 7 seconds. On our steady 128 BPM song, 7 seconds read 128.1 and 10 to 60 seconds read 128.0.