BPM Finder (Song BPM Detector)

What this BPM finder reads from your file

A song's tempo is simply the number of beats in one minute, and a BPM finder is a program that tries to find those beats without a human tapping along. This one takes up to 90 seconds from the middle part of the file, which is usually the most representative stretch: the intro is over, the drums have settled in, and the final fade has not started yet. Longer files are not slower to analyze, because only that window is used.

The result panel shows the main BPM, the same tempo read at half and at double speed, a tempo word such as moderato or allegro, and the length of one beat in milliseconds. It also shows the musical key, because the same pass over the audio gives a reasonable estimate of it. If you only care about tempo you can ignore the key; if you are planning a mix, the key sits right there next to the number. Once the analysis finishes, the drop zone folds away so the result has the whole page, and an Analyze another file button brings it back.

The search runs between 60 and 200 BPM. Music below 60 is usually felt at double that number and music above 200 at half, so the range covers the tempos people count along to. Inside that range the analysis prefers common tempos, which keeps it from reporting something odd like 173 when the real answer is 87 felt in two.

Everything happens in the page. The file is decoded on your device and analyzed there, and the audio is not sent to a server. WaveTinker does not fetch songs from streaming services; it works only with audio files you already have, so use it on tracks you own or have the right to use.

How the tempo is actually detected

The analysis first turns the audio into a curve that shows how strongly new sounds start at each moment. A kick drum, a snare hit or a plucked chord makes that curve jump; a held pad barely moves it. This curve is called an onset-strength envelope, and it is much easier to analyze than the raw waveform because it throws away pitch and keeps the rhythm.

Next, the program compares the curve with shifted copies of itself. If the pattern repeats every half second, the curve lines up well with a copy shifted by half a second, and the match score peaks there. A half-second spacing between repeats means two beats per second, which is 120 BPM. That comparison is called autocorrelation, and a comb of expected beat spacings between 60 and 200 BPM picks out the strongest candidate.

A beat-clarity score comes along with the answer. A track with a hard four-on-the-floor kick scores high; a solo piano ballad scores low. When the score is too low to trust, the page says "No steady beat" instead of showing a made-up number. That message is a useful result by itself: it tells you the track has no regular pulse for a program, or for a click track, to lock onto.

Every value on this page is an estimate. For a typical produced track with a steady grid the estimate is reliable, and for anything played by humans without a click it can drift or land on a neighboring interpretation. The sections below explain which cases to double-check.

Half time and double time: why 70 and 140 can both be right

Most BPM confusion comes from one fact: a beat is a matter of perception, not a physical object in the file. Take a drum pattern with a kick on one and three and a snare on two and four. Count along to the kick and snare together and you get 140. Count only the snare, with a relaxed head nod, and you get 70. Both readings describe the same audio, and both are correct in their own way.

That is why the result shows the main BPM together with its half and double. Trap and dubstep are the classic cases, along with half-time drum and bass. Producers write them at 140 or 170, but the feel is half of that, so listeners and DJs argue about which number is right. If a hip-hop track comes back at 140 and the genre table in the Numbers tab suggests 85 to 100, take the half value; a dance track that reads 64 probably wants the doubled 128.

A quick test settles it. Play the track and move your foot to the number shown. If your foot feels rushed, go to the half value. If it feels like wading through mud, go to the double. For most pop, house and rock the main reading is already the natural one, and you will not need either alternative.

When you plan to match two tracks, half and double matter in a practical way. A 70 BPM track and a 140 BPM track can be mixed without any speed change, because the beats line up every second beat. A 70 and a 128 cannot, and you would need to nudge one of them with the speed changer before they sit together. The result shows the half-time and double-time readings as chips; tap one to switch the number displayed. A Check by ear: song + click button plays the song with a click on every detected beat, so if the clicks drift away from the drums you know the tempo is wrong.

Songs that will not give a clean number

Live recordings and anything played without a click track are the hardest material. A drummer who pushes through a chorus and relaxes in a verse produces a tempo that wanders by several BPM, so there is no single true value, only an average. The finder reports the strongest repeating spacing in the analyzed window, and that is a fair summary. Do not expect two analyzers to agree to the decimal on a live track.

Rubato is the second troublemaker. Classical and jazz ballads often stretch and squeeze the pulse on purpose, so the beat stays flexible and the onset curve has no steady comb to match. Expect either a loose estimate or the message that no steady beat was found. That answer is correct: the music really does not have one.

Tracks with a long ambient intro can also confuse a reading. Because the finder uses the middle part, a song that starts with a 40 second drone and then kicks in at second 50 may be judged from a window that mixes both. If you know where the beat begins, trim the file first and analyze the trimmed version.

Electronic music with swing is a quieter case. A shuffled hi-hat pattern makes the onsets uneven, but the underlying kick still sits on the grid, so the tempo usually comes out right. What you may notice is a lower clarity than the same beat played straight.

A final case is a track with a tempo change, such as a song that speeds up in the bridge. A single number cannot describe it. Cut the file into sections with the trimmer and analyze each section separately if you need both tempos.

How to count BPM manually: the 15-second method

You do not need software to measure a tempo, and doing it once by hand makes every automatic reading easier to judge. Start a timer for 15 seconds, begin counting at the first beat you feel, and count every beat until the time is up. Multiply the count by four. If you counted 31 beats, the tempo is about 124 BPM.

The 15-second window is a compromise. Ten seconds multiplied by six works too, but the larger multiplier magnifies a miscount of one beat into a six BPM error. Thirty seconds multiplied by two is more accurate, but your attention drifts. Fifteen seconds with a factor of four keeps the error from a single missed beat to four BPM, which is good enough to tell 120 from 128.

Count the beat you would tap your foot to, not the fastest thing in the track. Hi-hats in sixteenths or a fast bassline are subdivisions of the beat, and counting them gives you four times the real tempo. If your answer lands above 200 or below 60, you have probably counted the wrong layer, so halve or double it.

To check the arithmetic backwards, divide 60,000 by your BPM to get the beat length in milliseconds. At 124 BPM that is about 484 ms. The page shows the beat length for the detected tempo, so you can compare your own count with the machine's in a few seconds.

Tap tempo as a second opinion

The best way to confirm a reading is to put a human in the loop. Play the track, then open tap tempo, which is one click away from the result, and tap along with the beat you actually feel. After eight or ten taps the average settles, and you can compare it with the detected value. If they agree, you are done. If tap says 70 and the finder says 140, you have just seen the half and double time effect on your own tapping.

Tapping is also the right tool when the finder reports no steady beat but you can hear one. A person follows a vocal line or a bass note that a program ignores, and averaging your taps gives a usable number for a live recording where an algorithm gives up.

The other way round works as well. If you tap a song from memory and are unsure about the result, load the file here and let the analysis check you. Agreement from two independent methods is stronger evidence than either one alone, and disagreement almost always means a factor of two rather than a real difference.

Putting the number to work

For a DJ set, the BPM tells you which tracks can follow each other with little or no pitch change. Tracks within a few percent of each other can usually be beat-matched by ear. Beyond that, the audio speed changer can bring a track to the target tempo, and the key shown next to the BPM tells you whether the pair will also sound good together. The key finder and Camelot wheel pages go deeper on that half of the job.

For a workout mix, the number is a pacing tool. Many runners like a cadence between 150 and 180 steps per minute, so a track at 160 or one at 80 felt in double time can match the stride. For video editing, knowing the beat length in milliseconds helps place cuts: at 120 BPM a beat is 500 ms, so a cut every two beats lands every second.

For practice, copy the tempo into the metronome and play along. The link from the result sets the metronome to the detected value, so you can check by ear that the click and the track agree, then slow the click down to learn a hard passage at a gentler speed.

Accuracy, formats and privacy

On clean, quantized music the reading is usually close, often within a BPM or two, and it is less reliable on loose performances. The result is shown with one decimal, such as 127.5. For production work where a tenth of a BPM matters, set the project tempo from the source of the track rather than from an analyzer.

If your browser can decode it, you can analyze it: MP3, WAV, M4A, AAC, OGG, Opus, FLAC and often the audio of a video. The size limit is 250 MB. Very long files are fine because only 90 seconds are analyzed, but decoding a huge file on an older phone can take a moment.

No upload happens at all. Your own device runs the analysis, and the file is not stored anywhere once you close the tab.