dB Toolkit

A sound level meter that keeps a record.

Most decibel apps show you a number. dB Toolkit gives you a dated report — the graph, the statistics, the loud events, and exactly what the reading is referenced to — for the times somebody else has to act on what you measured.

Is it for me? What it does

Three jobs, one document

The report is the same in each case. What changes is who reads it, and which published limit it is measured against.

A noise problem nobody is taking seriously

A neighbour, a pub, an extractor fan on a roof. Record the whole episode, not the loudest instant — when it happened and for how long is the part that carries weight. Measure the background too: BS 4142 and most other assessments are about the difference between a sound and the quiet it sits in.

For: a landlord, a council’s environmental health team, a managing agent, a solicitor.

A workplace you are responsible for

Daily exposure and peak, assessed against the Control of Noise at Work Regulations 2005, the OSHA permissible exposure limit and action level, or the NIOSH recommended limit. The app tracks the dose as you record and names the standard it used — OSHA and NIOSH can differ enormously on the same day of noise, so which one was applied is part of the answer.

For: a health and safety file, a risk assessment, an insurer, a union rep.

A venue or a show with a limit on it

Licence conditions and artist riders are often written as an A‑weighted average with a C‑weighted peak limit beside it. Log the show and leave with a timestamped record of what the level actually was — rather than an argument about it afterwards.

For: a venue manager, a licensing officer, a promoter, your own file.

In every one of them the useful thing is not the number. It is the number plus what it is referenced to, how uncertain it is, and what was going on when it was taken — which is what the report carries and a screenshot does not.

What it does

An instrument first, and a record keeper second.

The spectrum analyser full screen in landscape, showing a live narrowband trace, a peak-hold trace above it, the min/max envelope behind, and a crosshair labelled 1.3 kHz, E6, minus 42 cents
Turn the phone sideways and the analyser takes the whole screen. The crosshair names the loudest frequency — and the nearest musical note, with how far off it is.
The meter screen reading 95.2 dBA, labelled Damage risk, the arc well into the red, with a 30-second history graph showing the event and LAeq, LA10, LA50, LA90 and LAmax beneath
The meter, with the last 30 seconds and the statistics beside it. Loud events are marked as they happen.
The calibration settings: Decibel scale, measured for this device, plus 43.2 dB plus or minus 3 dB, with the reference microphone, serial number, date and method stated underneath
What the decibels are referenced to — the constant, its uncertainty, and how it was measured.
The exposure standard picker listing the UK lower and upper exposure action values, the UK exposure limit value, NIOSH REL, OSHA PEL and the OSHA action level, each with its source document and figures
Pick the standard to be assessed against. Each one names the document it comes from.
The signal generator: sine, square, triangle and sawtooth waveforms, a 1000 Hz frequency control and a level slider
Tones, noise and sweeps — and a warning that the meter can hear them.
A saved session playing back: the dB level graph above the audio waveform on one shared time axis, loud events marked in red, 81.3 dBA labelled on the graph, trim handles set to a six to fourteen second range
A saved session. The levels and the audio share one time axis, so you can trim to the part that matters — and the statistics follow the trim.
A preview of the PDF report showing the level graph, the audio waveform beneath it, a smoothed trend, the statistics row, the noise exposure figures and the peak events table
The report. Dated, with the scale it was measured on, every peak event, and any gap in the recording stated above the statistics.
The spectrogram: frequency up the side from 31.5 Hz to 16 kHz, time across, level as colour, showing sustained low-frequency energy and several bright transients
The spectrogram. When a noise started, how long it lasted, and what it was made of — which a live trace cannot show.
The analyser in third-octave mode, with bars per band and the lowest bands hatched rather than drawn as bars
Third-octave bands. The hatched ones are narrower than the analysis can resolve — drawn as unmeasured rather than as quiet.
The Pro screen listing what stays free - the meter, the weightings, the calibration, the statistics, the history graph - above an Unlock Pro button priced at 19.99 pounds
One payment. The meter, the weighting curves and the calibration are free and stay free.

Measurements you can quote

dBA, dBC and dBZ as real IEC 61672-1 filters, not labels. Fast, Slow and Impulse time weighting. LAeq, LA10, LA50, LA90, LAmax and Lpeak.

Honest about accuracy

iPhones do not publish a microphone sensitivity, so most sound level apps show a number referenced to nothing at all. This one ships a constant measured against a calibrated reference microphone, and states its uncertainty on every report.

The record

Record a session of any length — it keeps measuring with the screen locked. The audio is saved too, and every interruption is stated in the report, because a gap nobody mentions is worse than no measurement.

Measured against a standard

Pick a published limit — the Control of Noise at Work Regulations 2005, BS 4142, BS 8233, OSHA or NIOSH — and get a verdict in the report, with the source document named.

Noise exposure

Cumulative dose while you record and in the report, against the standard you choose. OSHA and NIOSH can differ enormously on the same day of noise — eight hours at 95 dB(A) is 200% of the OSHA permissible dose and about 1008% of the NIOSH one — so the app always names which one it used.

Analyser and generator

A real-time spectrum analyser — narrowband, octave and third-octave, with averaging, peak hold and a min/max envelope — and a signal generator with tones, white, pink and brown noise, and sweeps.

The spectrogram

The spectrum over time, so you can see when a noise started and how long it lasted rather than only what it is made of now. Feedback building before it rings is a line thickening; a room mode is a horizontal stripe; a neighbour’s bass is a band that starts and stops.

Room measurement

Play a sweep, record it back, and get the reverberation time per octave band, the clarity figures and the arrival time. Where a band’s decay disappears into the noise it returns nothing rather than inventing a figure, and says so.

A verdict with three answers, not two

A phone measures on a scale that carries a stated uncertainty — three decibels on the one iPhone measured against a reference microphone, five on every other, and the app says which you have. Against an 85 dB limit that makes any reading between roughly 82 and 88 genuinely undecidable, or 80 and 90 on an estimated scale — and a workplace at 86 dB is exactly the case somebody needs an answer about.

So the app answers pass, fail, or too close to call. The third one is not a hedge. A confident tick there would be a made-up certainty in a document you might act on.

Fully offline

No account, no analytics, no advertising, no server. The app makes no network requests at all. Your recordings never leave your device unless you share them yourself.

One payment, no subscription, and no ads — an app that plays advertising audio while measuring is contaminating its own reading, which is the whole reason this one exists.

What it is not

This is a phone, not a Class 1 or Class 2 sound level meter. For a measurement that has to stand up on its own in court or in a formal enforcement action, you need a calibrated instrument and somebody qualified to use it.

What this gives you is a competent, honest, timestamped record with its uncertainty stated — which is very often the thing that is actually missing when somebody is trying to get a noise problem taken seriously.