Infrasound Monitor
Watch a live sub-20 Hz spectrum and a slow time-domain waveform for long-period signals, log the dominant low-frequency peaks, and match them against a guide of published infrasound source bands — road and rail traffic, wind turbines, large HVAC fans, weather and geological activity.
⚠ This is an experimental, device-limited view — not a real infrasound measurement. Consumer microphones roll off steeply below ~50–100 Hz, and the OS/browser high-pass filter plus auto-gain typically remove almost everything below 20 Hz. So whatever appears in the sub-20 Hz band here is heavily attenuated and largely unreliable — treat it as a curiosity, not data. Genuine infrasound monitoring needs a calibrated infrasound microphone / microbarometer (or an accelerometer for structure-borne motion). The decibel readings are relative dBFS, not calibrated SPL, and the source guide is a published-typical reference, not a diagnosis. Auto-gain, noise suppression and echo cancellation are requested off; nothing is recorded or uploaded.
Idle — press Start to allow your microphone and watch the sub-20 Hz band.
Peaks below −70 dBFS are ignored in the log.
Vertical axis = relative level (dBFS, uncalibrated) — not SPL. The slow waveform shows a coarse trace of the microphone’s low-frequency/DC-ward level over the last several seconds (a long-period indicator, not a true waveform). Dashed magenta = Schumann reference (7.83 Hz, electromagnetic — shown for orientation only).
Live reading
A 32768-point FFT is used for the finest low-frequency resolution the Web Audio API allows. Even so, a meaningful read below ~50 Hz depends entirely on your hardware — most built-in mics simply do not pass it.
Peak log
The dominant sub-20 Hz peak is logged at most once per second when it rises above your sensitivity floor (newest first, capped at 60 rows). Each row carries the browser’s local timestamp, the peak frequency, its relative level, and the closest published source band.
| Time | Peak | Level | Closest band |
|---|---|---|---|
| Start monitoring to log infrasound peaks. | |||
Infrasound source reference (published-typical bands)
These are published-typical frequency bands, not exact constants — real sources vary with size, speed and conditions, and several overlap. Use them to interpret a peak, not to confirm a source. Sources are cited in the explanatory sections below.
| Source | Typical band | Notes |
|---|---|---|
| Ocean waves / microbaroms | ~0.1–0.5 Hz (peak ~0.2 Hz) | Standing-wave ocean–atmosphere coupling; the dominant natural infrasound on global IMS arrays. |
| Severe weather / tornadoes | ~0.5–5 Hz | Atmospheric pressure disturbances detectable over long ranges. |
| Wind turbines (blade pass) | ~0.7–1.5 Hz fundamental + harmonics | Blade-pass = blades × rotor speed; harmonics extend upward into low-frequency noise. |
| Heavy road traffic (truck suspension) | ~3 Hz (body) & ~10–16 Hz (wheel/tire) | Suspension & tire resonances of heavy vehicles; rail traffic adds low-Hz tones. |
| Volcanic / explosion / sonic boom | ~0.1–10 Hz (often 1–3 Hz transient) | Broadband pressure waves; monitored by the CTBTO IMS infrasound network. |
| Schumann resonance | 7.83 Hz (electromagnetic) | Earth–ionosphere cavity resonance — not acoustic; shown only as a frequency landmark. |
| Large HVAC fans / machinery | ~5–16 Hz tonal | Big ventilation fans & rotating plant radiate low-frequency tones, often structure-borne. |
| Elephant rumbles (for reference) | ~15–35 Hz | Biological infrasonic communication; the fundamental of these calls dips below 20 Hz, with harmonics extending upward into the audible range. |
How to Use
1. Set realistic expectations first. Before you start, understand that an ordinary laptop or phone microphone almost certainly cannot capture true infrasound. This monitor is a learning and exploration tool: it shows you the very bottom of what your hardware passes and lets you reason about source bands. If you need to actually measure infrasound, you need a calibrated infrasound microphone or microbarometer.
2. Pick a microphone and press Start. Grant microphone permission when prompted. The tool requests the raw signal with auto-gain, noise suppression and echo cancellation switched off so those features do not reshape the low end further. The sample rate and the resulting Nyquist limit are shown in the Live reading panel.
3. Choose a spectrum range and averaging. The default zoom shows 0.5–20 Hz; the extended option reaches 30 Hz to include the infrasound–to–audible transition. Heavier averaging steadies the trace so a persistent low tone stands out from random fluctuation; Off shows the instantaneous spectrum.
4. Watch the slow waveform. Infrasound is felt as slow pressure swings rather than heard. The slow-waveform panel plots a coarse, low-pass-smoothed trace of the microphone’s low-frequency level over a multi-second window, so slow drifts are visible even when the spectrum bins are noisy.
5. Read the peak log and source guide. When a sub-20 Hz peak rises above your sensitivity floor it is logged with a timestamp and matched to the closest published source band. Use the reference table to interpret what a peak might be — never as proof. Export the log as CSV for your own notes; it is a local download and nothing leaves your device.
Understanding Your Results
Why sub-20 Hz reads are unreliable on consumer gear
The single most important thing to understand is that the bottleneck is your microphone, not the software. MEMS microphones in phones and laptops, and the OS/browser audio pipeline behind them, apply a high-pass filter that rolls off steeply below roughly 50–100 Hz, and automatic gain control further distorts very low frequencies. By the time you reach 20 Hz and below, almost nothing genuine is left. Anything the spectrum shows down there is dominated by the device noise floor, handling vibration, and filter artefacts. That is why this tool frames the live sub-20 Hz view as experimental and device-limited.
The Nyquist limit (an upper bound, not your problem here)
Your browser’s AudioContext runs at a fixed sample rate, typically 44,100 or 48,000 Hz, which sets a hard Nyquist limit of about 22,050 or 24,000 Hz — the highest frequency that can ever be captured. The tool reads this at runtime and displays it. For infrasound the Nyquist ceiling is irrelevant; the limiting factor is the steep low-frequency roll-off described above.
Levels are relative, not calibrated
The decibel numbers are dBFS — relative to digital full scale — not calibrated sound pressure level in dB SPL. They are useful for comparing one moment to another with the same mic and gain, but they cannot tell you the true acoustic level of an infrasound source. Calibrated infrasound measurement requires a sensor with a known sensitivity curve.
The source guide is interpretation, not identification
The closest-band match is a convenience based on published frequency ranges. Many sources overlap (a 3 Hz peak could be truck suspension, severe weather, or a building mode), and a single number cannot distinguish them. Real source identification uses arrays of calibrated sensors, multiple stations, and context — exactly what global networks such as the CTBTO International Monitoring System do.
How It Works
The monitor streams audio from your microphone into a Web Audio AnalyserNode with a 32768-point FFT — the largest the API allows — to get the finest possible frequency resolution at the low end. At a 48 kHz sample rate each bin is about 1.46 Hz wide (48000 ÷ 32768), so the 0.5–20 Hz band is covered by only a handful of bins. The spectrum panel stretches those bins across the canvas; an exponential moving average smooths the trace when averaging is enabled. Bin 0 (DC) is skipped so a constant microphone offset is never mistaken for a real tone.
The slow waveform collapses each analysis frame to a single average (DC-ward) value, one-pole low-pass smooths that running level, and accumulates a multi-second rolling buffer (bounded in length) so slow drifts in the microphone’s low-frequency level are visible. It is a coarse long-period indicator rather than a true band-limited pressure waveform. Because infrasound is experienced as pressure changes over time rather than as audible pitch, this long-period view can be more intuitive than the spectrum.
Health and perception reference (published-typical, not medical advice)
Human hearing sensitivity falls off dramatically below 20 Hz. In the classic measurements summarised by Møller & Pedersen (2004, Noise & Health) and Watanabe & Møller (1990), the median hearing threshold rises from around 88 dB SPL at 16 Hz to roughly 107 dB SPL at 4 Hz, and well above that toward 2 Hz — meaning a sound must be extremely loud before it is perceived at all at these frequencies, and below about 10 Hz it tends to be experienced as a sensation of pressure rather than as a tone. The dynamic range is also compressed: a few decibels’ change at 4 Hz subjectively resembles a much larger change at 1 kHz. High-level infrasound (well above these thresholds) has been associated with annoyance and sensations of vibration, but a level below the threshold of perception is neither heard nor felt. These figures are published references for orientation — this tool is educational and does not perform any clinical or diagnostic hearing assessment, and nothing here is medical advice.
What this tool can and cannot do
It can teach you what the infrasound band looks like, show the limits of your own hardware, and help you reason about candidate sources. It cannot deliver a calibrated, trustworthy sub-20 Hz measurement on a consumer microphone. For genuine work — wind-turbine compliance, building diagnostics, volcanic or atmospheric monitoring — the right instruments are a calibrated infrasound microphone, a microbarometer, or an accelerometer for structure-borne motion.