How loud is too loud: a dB(A) table with the hours you can stay

“Is this too loud?” is not really a question about decibels. It is a question about decibels and time. 85 dB(A) for eight hours and 100 dB(A) for fifteen minutes do roughly the same damage. That trade is the whole story, and it is why a single number on a screen is only half an answer.

→ NoiseLab measures dB(A) and shows the hours you have left, free on Android

The table

Levels are A-weighted (dB(A)), the weighting that follows how human hearing responds. Hours are the daily exposure limit from EU directive 2003/10/EC: 85 dB(A) over 8 hours, with a 3 dB exchange rate, which means every 3 dB louder halves the time.

Level	Sounds like	You can stay
30 dB(A)	quiet bedroom at night	all day
40 dB(A)	library, fridge hum	all day
50 dB(A)	quiet office, rain	all day
60 dB(A)	normal conversation at a metre	all day
70 dB(A)	busy restaurant, vacuum a room away	all day (WHO suggests keeping the daily average here)
80 dB(A)	busy street, alarm clock	16 h (below the EU action limit)
85 dB(A)	heavy traffic, blender	8 h
88 dB(A)	motorbike passing	4 h
91 dB(A)	lawnmower, subway platform	2 h
94 dB(A)	club floor, kids party in a hard room	1 h
97 dB(A)	chainsaw at a distance	30 min
100 dB(A)	loud rehearsal, power tools	15 min
103 dB(A)	front rows at a gig	7 min
106 dB(A)	drum kit at a metre	4 min
112 dB(A)	siren nearby	under a minute
120 dB(A)	jet at 100 m, threshold of pain approaching	avoid

Two things this table does not say, and both matter:

• It is a daily budget, not a wall. Fifteen minutes at 100 dB(A) plus an hour at 94 dB(A) is already two full days’ worth. Exposure adds up across the day.
• Peaks are separate. A single hammer blow or snare crack can pass 130 dB peak without moving the average much. The EU sets a peak limit at 137 dB(C); a proper meter reports the unweighted peak alongside the average for exactly this reason.

Which number to look at

Sound level meters report several things, and people usually read the wrong one.

• Fast / Slow: the level right now, with a fast (125 ms) or slow (1 s) time weighting. Good for pointing at a source. Useless for “is this room too loud today”.
• LAeq: the A-weighted average over the whole measurement, energy-weighted. This is the number the exposure rule is written against, and the one to screenshot.
• Lmax / peak: the loudest moment, and the unweighted peak. For bangs and impacts.
• L10 / L50 / L90: the level exceeded 10%, 50% and 90% of the time. The gap between L10 and L90 tells you whether it is a steady hum (small gap) or a series of bangs (large gap). L90 is also the practical way to read a room’s noise floor.

If you take one habit from this article: judge rooms by LAeq over a few minutes, not by the number that flickers when you walk past a fan.

Measuring it with a phone, honestly

A phone can do this, within limits you should know:

• A phone does not know how loud a pascal is. It reports level relative to full scale, and turning that into dB SPL needs a calibration constant. Uncalibrated, expect around 5 dB of error. Five dB is the difference between “8 hours” and “2.5 hours”, so it matters. A meter that never mentions this is hiding something.
• Calibrate once against a real sound level meter or a 94 dB calibrator: read both, type what the reference says, and the offset sticks with that phone.
• Automatic gain control ruins it. The measurement has to use the raw, unprocessed microphone input, or the phone will quietly compress exactly the loud parts you are trying to measure.
• The honest range is roughly 30 to 105 dB(A). Below that you are measuring the mic’s own noise; above it the mic clips and the reading flattens. A good app tells you when you are out of range instead of printing a confident number.
• Weighting matters. A for how ears respond and for exposure rules, C for peaks and low frequency content, Z (flat) when you want raw physics.

What to do with the reading

• Home and office: aim for a daily average under 70 dB(A). Above that it is fatigue rather than damage, but fatigue is real.
• Rehearsals and gigs: measure once with the band playing, then read the hours. Most rooms come out between 95 and 105 dB(A), which means 15 to 60 minutes. That is an argument for earplugs, not for stopping.
• Kids: the same rules apply with less margin. Parties in hard rooms hit 90 to 95 dB(A) easily.
• A complaint: bring a timeline, not a shout. A measurement showing L90 and the peaks over half an hour is a document; “it was really loud” is not.

Measured with NoiseLab: LAeq, Fast and Slow, peaks, L10/L50/L90, A/C/Z weighting, the hours you have left, and a line on every screen when the reading is uncalibrated. Free to measure on Android, no ads, no subscription.