The acoustic stud finder that listens. Wallrap uses your iPhone's microphone to classify each wall knock as hollow, edge, or solid in real time, then turns saved traces into Mount Check reports with confidence, warnings, and verification steps. Fully offline.
Carpenters find studs by tapping the wall and listening for the pitch shift — drywall over a cavity sounds lower and longer than drywall over a stud. Wallrap runs a 1024-sample real-FFT on each knock (via Accelerate vDSP) and reduces it to two numbers: the spectral centroid and the low-band-energy ratio (50–500 Hz). A two-axis classifier then maps every knock onto a hollow→solid axis with a soft confidence score.
Tap a spot you know is hollow, then a spot you know is solid. Wallrap captures the acoustic signatures of your specific wall — material, thickness, backing — and uses those references as the classifier boundaries. The factory defaults work, but calibrating for your wall makes the readings markedly better.
Tap the wall and instantly see a HOLLOW, EDGE, or SOLID chip with a soft confidence ring. The live spectrum and waveform display update with each knock so you can see what the microphone heard.
Slide the phone left-to-right along the wall, knocking every inch or two. The trace strip builds a color-coded map of hollow (blue) and solid (amber) zones — a quick visual picture of where the framing is.
Two-step calibration captures your hollow and solid reference knocks. The classifier uses the midpoint between your references as the decision boundary, weighted by both the spectral centroid and the low-band-energy ratio.
The haptic feedback is tuned to the classification: a crisp, strong impact for solid (stud); a softer, lower thump for hollow (cavity). Intensity is adjustable. Works on all iPhones with a Taptic Engine.
Open a saved trace to see the strongest candidate anchor point, confidence score, spacing evidence, drill-risk warnings, and a practical checklist before you mark the wall or hang an object.
In Listen mode, the last 3–5 knocks are tracked in a rolling window. When they all agree with high confidence, a "STUD FOUND" or "CAVITY CONFIRMED" banner appears — far more decisive than watching individual readings flicker. A row of colored dots shows the running sequence at a glance.
After a trace, Wallrap analyses the gap pattern between solid segments to infer whether the wall is framed 16" on-center (the US residential standard per IRC R602.3) or 24" on-center. A confidence bar and a one-line summary appear beneath the trace strip. The predictor is honest about uncertainty — it shows "?" when the data isn't clear enough.
Completed traces can be saved with an optional room note. The History tab shows every saved session with its wall map, knock counts, and spacing estimate. Tap any session to review it or export the data.
Each knock is captured via AVAudioEngine, windowed by a Hann function, and transformed via a 1024-point real-FFT (Accelerate vDSP). The spectral centroid and 50–500 Hz low-band-energy ratio are extracted per-knock. A two-channel linear classifier maps these features onto the hollow→solid axis using the stored calibration profile. Knock onset detection uses an EMA-smoothed noise floor with a configurable threshold and 80 ms debounce gate. All DSP runs on-device in real time with no audio recording.
Wallrap is an acoustic estimator, not a certified stud finder. Noisy environments, inconsistent knocking force, thick walls, plaster-on-lath, and metal studs all reduce accuracy. Always confirm stud positions with a pilot hole or a hardware stud finder before driving large fasteners. The app includes this disclosure on the Calibrate screen and in every relevant context.
No account. No ads. No analytics. No network access. No in-app purchases. No subscriptions. Calibration data and saved sessions are stored locally in UserDefaults. Microphone access is used only while you are actively listening, tracing, or calibrating — never in the background.