Point your phone's mic at a speaker or tap along — get an accurate tempo reading in seconds, with the mobile app.
Whichever fits the moment — the same neural-net engine under the hood.
Point your phone's mic at a speaker, a band, a DJ set, or an instrument. The detector locks onto the beat and reports the BPM with decimal precision.
Tap along to a beat in your head and get an instant average. Auto-resets if you pause, so you can keep refining your estimate.
A live chart tracks every reading over the last few minutes — perfect for spotting tempo drift in a rehearsal or a DJ set.
A short pipeline — every step runs on your phone.
Your phone's mic feeds a rolling 12-second buffer at the device's native sample rate.
Resample to 11.025 kHz, take the log-mel spectrogram (256 frames × 40 bins).
A compact AI model outputs a 256-class softmax over BPM values from 30 to 285.
Parabolic interpolation around the peak gives a decimal BPM reading — not just the nearest integer.
All inference happens locally on your device using TensorFlow. There's no server-side audio processing, no recording, no account required.