Beat Tracking: How Computers Find the Pulse in Music
Beat tracking is the task of finding the regular pulse in a piece of music, the points where a listener would naturally tap a foot. Software does it in two stages. First it detects note onsets, the moments when sounds begin. Then it searches for a steady pattern of time intervals that fits those onsets and follows it through the piece, adjusting when the tempo drifts. Modern systems use neural networks trained on recordings with human beat annotations. Pianists can borrow the same idea: listen for where notes begin, find the repeating gap and keep it steady in the body.
Why tapping along is harder than it looks
Most people can tap along to a pop song without thinking. For a computer, the task is surprisingly tricky. A recording does not contain a label that says "beat here". It contains a mix of instruments, some playing on the beat, some between beats and some holding long notes. Classical recordings may slow down and speed up freely. Jazz may place accents off the beat on purpose. The software has to infer a pulse that a human listener would agree with.
Step one: onset detection
An onset is the start of a note. Drums and piano notes have sharp onsets; strings and voices often fade in more softly. Onset detection looks for sudden increases in energy or changes in the frequency content of the sound. The result is an "onset strength" curve that rises whenever something new begins.
Why piano is easier
Each piano note starts with a hammer strike, so its onsets are clear. Bowed or sung music is harder because notes can begin very gradually.
Step two: finding the period
Next, the system looks for regular spacing in the onset curve. A common method, called autocorrelation, compares the curve with time-shifted copies of itself. If the curve lines up well when shifted by half a second, the music probably has a beat every half second, or 120 beats per minute.
- Tempo estimate: how far apart the beats are.
- Phase estimate: where exactly each beat falls.
- Tracking: adjusting both as the music moves.
Some systems also estimate the downbeat, the first beat of each bar, which makes it possible to count in bars rather than just beats.
The double-or-half problem
A frequent mistake is choosing a tempo that is twice or half what listeners would tap. A song at 70 beats per minute might be labelled 140 if the hi-hat plays fast notes. Human listeners face the same ambiguity, and different people may tap at different levels. Researchers often accept both answers when evaluating systems, because the "correct" pulse can depend on the listener.
Learning from human annotations
Neural network beat trackers are trained on many recordings where people have marked each beat by hand. The network learns which sounds tend to coincide with beats in different styles. Combined with a model that prefers steady tempos, this approach handles a wide range of music, though very free rubato or unusual meters can still confuse it.
Where beat tracking is used
- DJ software that matches the tempo of two tracks.
- Music games that time visual cues to a song.
- Lighting systems that flash on the beat at live shows.
- Practice tools that measure whether a player stays steady.
For piano learners, the last use matters most. A practice tool that knows where the beat should be can show when you rushed or dragged.
Training your internal beat tracker
You can practise the same steps in your own listening.
Find onsets. Listen to a recording and notice every new note in the bass line.
Find the period. Tap along until your taps feel even, then count aloud.
Track changes. Choose a piece with a gentle ritardando and try to follow it with your taps without stopping.
Play against it. Put on a recording, find the pulse, then play a simple chord on each beat. Record yourself and check whether your chords stay with the music.
FAQ
Q: What is the difference between tempo and beat?
A: The beat is the regular pulse; the tempo is how fast those pulses come, usually measured in beats per minute.
Q: Why does my app show double the tempo I expect?
A: Fast subdivisions such as hi-hats can make the software choose a quicker pulse level. Both levels can be musically valid.
Q: Can beat tracking follow classical rubato?
A: Partly. Gentle tempo changes are often tracked, but very free timing remains difficult.
Q: How can I improve my own sense of pulse?
A: Tap along to many styles, count aloud, and record yourself playing with and without a metronome.
Keep a steady pulse with TIGA ALPHA
TIGA ALPHA is an AI piano teaching app built by a Thai team, with lessons in Thai, English and Mandarin. Its CEO, Teacher Aum, has 20 years of experience in piano teaching, and that music teaching background shapes every lesson, while piano teaching with AI gives you feedback each time you play piano. Use it as an online piano lessons app for short daily sessions, follow it as a complete AI online piano course, or combine it with a piano course or piano learning course led by a teacher. Whether you want to learn piano from zero, learn piano online after a long break, learn music alongside your children or simply play music for fun, this online piano teaching application keeps your next step clear. New users get a 7-day free Premium trial, then 799 THB/month. Open the piano teaching app below and pick a first lesson that fits your week.
- Download and start: TIGA ALPHA landing page
- In-person lessons in Thailand: TIGA Studio
- About the company: tigaalpha.com
A steady beat starts as a feeling in the body and grows with every focused practice session.