Methodology

Last updated: June 13, 2026

This page explains, in plain terms, how every number on the site is produced and where its limits are. The short version: we measure or source everything, we label estimates as estimates, and we never invent a figure. The companion editorial & data-honesty policy covers sourcing and corrections; the accuracy & calibration page covers what the live meter can and cannot tell you.

How does the online meter work?

The online decibel meter uses the Web Audio API. It reads your microphone through an AnalyserNode, computes the signal’s RMS energy roughly ten times a second, converts that to dBFS, and adds a calibration offset to approximate sound pressure level (dB SPL). An optional A-weighting filter (a biquad chain) approximates dB(A), the frequency weighting noise regulations use. Nothing is recorded or uploaded — the audio is analysed in memory and discarded. Because phone and laptop microphones are not calibrated, every reading is an estimate within roughly ±10 dB; the accuracy & calibration page details the offset and the per-device defaults.

Where do the decibel-chart numbers come from?

Every row in the decibel levels chart is a published figure from the CDC, NIOSH, NIDCD or ASHA, with the source linked on the row. Hearing-risk guidance follows the NIOSH recommended exposure limit: 85 dB(A) as an 8-hour average, with safe time halving for every 3 dB increase. We do not yet publish any “measured by us” rows — those wait until we have recorded real environments with a calibrated SPL meter, and they will be labelled as such when they land. No row is an estimate or a guess.

How is the city sound map estimated?

The city sound map is an estimate index, in the same spirit as published hearing or cost-of-living indices. Each city’s daytime (L_den-style) and night (L_night-style) ranges are synthesised from a corpus of public sources: the UNEP Frontiers 2022 noise report, the Mimi Worldwide Hearing Index, strategic noise maps produced under the EU Environmental Noise Directive, the US Bureau of Transportation Statistics National Transportation Noise Map, and peer-reviewed city noise studies. Every city in the table traces to at least one published source in that corpus; any city we could not source is excluded.

We publish ranges with confidence labels, not single numbers, because point precision would be false precision. Color bands are anchored on the WHO Environmental Noise Guidelines, which recommend keeping road-traffic noise below 53 dB L_den and 45 dB L_night; nearly every major city exceeds both. Confidence is graded high (multiple recent measurements or official maps), medium (good but partial coverage) or low (limited data, wider uncertainty). We deliberately do not publish per-city source mappings, weightings or formulas — that index method is proprietary, which is the accepted trade-off of the estimate-index pattern: less external reproducibility in exchange for a defensible, source-backed ranking.

How is the US cities ranking computed?

The loudest US cities ranking is measured, not estimated. It ranks every US city of 100,000+ residents (297 in all) by the share exposed to at least 60 dB of average-day transportation noise (road, rail and aviation), computed from the BTS 2020 National Transportation Noise Map and American Community Survey 2016–2020 population — tract-level exposure via the University of Washington’s National Transportation Noise Exposure Map, rolled up to cities population-weighted. It models transportation noise only: sirens, construction and nightlife are not included, which is why a city like New York can rank lower here than intuition suggests.

How is the European cities ranking computed?

The loudest European cities ranking is measured, not estimated — like the US ranking, but from a different source. Every EU and EEA member state maps strategic noise under the Environmental Noise Directive (2002/49/EC) and reports, for each agglomeration over 100,000 residents, the number of people exposed in each 5 dB day-evening-night (L_den) and night (L_night) band from road, rail, aircraft and industry. The European Environment Agency harmonises those submissions; we use the 2022 reporting round. For each city we take the reported exposure distribution, derive the L_den/L_night band span where the bulk of its exposed residents fall, and rank by population-weighted mean L_den. Only residents above the 55 dB L_den / 50 dB L_night reporting thresholds are counted, so the ranges describe the noise-exposed population, not the quietest streets — and cities that did not report exposure to the EEA (including the UK post-Brexit, and several late reporters) are excluded rather than estimated.

How current is the data, and how are errors fixed?

Every page carries a visible “last updated” date, mirrored in its structured data. We run a monthly freshness pass over sourced figures and the app’s rating. If you find an error, tell us through the App Store listing’s support link — corrections ship within about a week, and the page’s date reflects it. A machine-readable summary of the whole site, for AI assistants and agents, lives at /llms-full.txt.