Drop in an MP3 and get the beat grouping, the tempo, and a beat grid you can check against the track. Your file is analysed on your own device and never uploaded.
Runs on the Web Audio API. Your file never leaves your device. The first 90 seconds are analysed.
Audio tells you how beats group together. It cannot tell you how a musician chose to write that grouping down — so the result names both readings rather than picking one.
| Result | What it means | Typical of |
|---|---|---|
| 4/4 | Beats group in twos and fours. The most common grouping by a wide margin. | |
| 3/4 · 6/8 | Beats group in threes. Telling 3/4 from 6/8 needs a second level of pulse analysis this tool doesn't do, so both readings are shown. | |
| 2/4 noted | The bar appears to be felt in two rather than four. Often written either way for the same music. | |
| No clear grouping | A steady beat was found but no repeating accent pattern stood out. Rubato, sparse arrangements and heavily compressed masters all do this. |
Any MP3, WAV, FLAC or M4A. A section with steady drums works best — an intro or a breakdown gives the analysis less to lock onto.
Onsets are tracked across three frequency bands to place the beats, then the loudness of each beat is compared to find which ones are accented.
The grid shows the repeating accent it locked onto. Count along with the track — if the tall bars land where you feel the strong beats, the reading is right.
Because this tool doesn't measure the thing that separates them. In principle they differ: 3/4 has three main beats per bar, while 6/8 has two main beats that each split into three. Hearing that apart means analysing two levels of pulse at once — the beat and how it subdivides. This tool tracks a single pulse and asks how those beats group, which tells you the music is in threes but not which level you're counting. So it reports both readings rather than picking one it hasn't actually tested. As a rule of thumb you can apply yourself: if the bar feels like a long-short lilt in two, it's 6/8; if you naturally count three even beats, it's 3/4.
The analysis needs beats whose loudness actually varies in a repeating pattern. If a track is rubato, very sparse, or mastered so hard that every beat hits the same level, there is no accent pattern left to measure. It reports that instead of falling back on 4/4 — since most music is in 4, defaulting to 4 would look correct most of the time while telling you nothing.
Measured two ways. On synthetic drum patterns with known groupings it gets all 52 test cases right, including refusing to answer on silence, held tones and clips that are too short. On a set of 12 real full-mix tracks across hip-hop, house, funk, synthwave, reggaeton, lo-fi, indie, country, jazz, K-pop, R&B and cinematic, it answered 8 and every one of those 8 was correct — the other 4 came back as no clear grouping rather than a wrong guess. So it stays silent more often on real recordings than on clean drum loops, and dense percussion helps. There is one failure mode worth knowing about: a slow piece in 6/8 whose only accents fall on the two main beats, with nothing playing on the subdivisions in between, gives this tool a plain strong-weak alternation to look at — the same thing a two-feel produces. It can call that 4/4 with high confidence and be wrong. If you are working on a sparse 6/8 ballad, check the answer against your own count.
Not currently. It tests groupings of 2, 3, 4 and 6, which covers the overwhelming majority of recorded music. An odd-meter track will usually come back as no clear grouping rather than a wrong answer.
It tells you how many beats are in a bar and which note value counts as one beat. The top number is the count — in 4/4 you count to four and start again, in 3/4 you count to three. The bottom number says which note gets the beat, which matters for reading sheet music but not for hearing the pulse.
No. Decoding and analysis both run in your browser using the Web Audio API. Nothing about the file leaves your device, and there is no account to create.
Tempo has a genuine octave ambiguity — a track at 170 BPM counted in twos is the same waveform as 85 BPM counted in ones. The analysis leans toward readings near 120 BPM because that's where most music sits, so fast tracks occasionally come back at half speed. The grouping is usually still right; use the BPM Detector if you need the tempo on its own.
Upload a clip and read the beat grid — it all runs in your browser.
One place to generate, convert, detect, and remix.
Turn a prompt into a full song with vocals.
Write full song lyrics from a theme or a line.
AI rap diss with lyrics + beat in seconds.
Convert any audio into editable MIDI notes.
Add a slowed + reverb tape vibe, free in your browser.
Find any track's tempo by tap or upload.
Detect the musical key of any song.
Check your pitch accuracy note by note.