The Decibel, Demystified

A decibel is not a unit of loudness. It is a ratio between two quantities on a logarithmic scale, so every "dB" printed on a fader, a spec sheet or a meter is a comparison against something else. Find out what it is being compared to and those numbers stop being decoration.

How Sound Actually Works left off at why the unit exists: amplitude covers an enormous range and your ears do not respond to it in a straight line. What musicians get wrong is what comes next. They treat dB as a fixed unit like a volt, so two numbers with different letters after them look subtractable — and a keyboard patched into a mixing desk turns up mysteriously quiet with nothing obviously broken.

Your Hearing Covers a Million to One

The quietest pressure swing a healthy young ear can detect in the midrange is about 20 micropascals, twenty millionths of a pascal, and sound turns physically painful somewhere around 20 pascals. That is a million to one in pressure, and because power goes as the square of pressure, a million million to one in power.

A logarithm turns multiplication into addition, so every tenfold jump in power becomes the same fixed step and the whole audible range fits between roughly 0 and 120. That is more than tidiness: your hearing responds to ratios rather than differences, so a scale built on ratios tracks perception better than raw pressure does. Going from 1 watt to 2 watts on a guitar amp and going from 50 watts to 100 watts are the same size step to your ear — and the same number on this scale.

A Decibel With No Reference Is Not a Number

One word, two jobs. Most of the time a decibel is describing a change: cut 3 dB at 400Hz, this cabinet is 6 dB louder than that one. Nothing needs referencing there, because both things being compared are in front of you.

The other use is an absolute level, and it means something only once someone tells you where zero sits. A powered speaker whose datasheet lists its input sensitivity as "−2 dB" is describing a voltage it has declined to name — the figure is unusable until you know what it is two decibels below. Attach a letter and the same number becomes a real claim about a real signal, because the letter pins zero to a stated physical quantity. Two specs carrying different letters are not comparable until you convert one.

The Letter After "dB" Is the Whole Point

Four references cover nearly everything you will run into:

• dBu — 0.775 volts RMS. The odd number is a fossil of telephone practice: one milliwatt into a 600-ohm line works out to 0.7746 volts. The impedance convention went away and the voltage stayed. Professional line-level gear uses it.
• dBV — 1.0 volt RMS, a round number rather than a physical landmark. Consumer equipment and RCA connections use it.
• dB SPL — 20 micropascals of sound pressure in air, roughly the threshold of hearing. This is what a meter or phone app at the mix position reads.
• dBFS — digital full scale, the largest number the format can hold. Zero is the ceiling rather than the middle, so anything short of full scale reads negative.

Professional gear runs a nominal +4 dBu and consumer gear runs −10 dBV, and the difference is not 14 dB. That subtraction is the most repeated error in the subject, and it fails because the two are counted from different zeros. Convert both to volts and it comes apart cleanly: +4 dBu is 1.23 volts, −10 dBV is 0.32 volts, a ratio a little under four to one, which is 11.8 dB. Call it 12.

Feed a laptop or a consumer keyboard into an input expecting +4 dBu and the signal lands around 12 dB shy, so you make it up on the preamp and haul the noise floor along with it. Go the other way and a professional output can push a consumer input into distortion. Neither box is faulty — they were built to different zeros.

Three Is Power, Six Is Voltage, Ten Is an Opinion

There are two equations, and picking the wrong one is where most dB arithmetic falls over. For power, meaning watts electrical or acoustic, a decibel is ten times the base-ten logarithm of the ratio between the two values. For anything amplitude-like, such as volts or sound pressure, the multiplier is twenty instead, because power rises with the square of amplitude.

• Power — dB = 10 × log₁₀(P₂ ÷ P₁), for watts.
• Amplitude — dB = 20 × log₁₀(V₂ ÷ V₁), for volts, current or sound pressure.

Three consequences are worth memorising. Plus 3 dB is double the power: two identical amplifiers instead of one, or 500 watts where you had 250. Plus 6 dB is double the voltage or double the sound pressure, so a microphone output that doubles has gone up 6 dB, not 3. Plus 10 dB is about twice as loud, and that third claim is a different kind of statement from the first two.

The first two are arithmetic and a meter will confirm them, while the third comes from asking people: listeners told to set one sound to half the loudness of another land roughly 10 dB down, in the midrange, at ordinary levels. That is an average of opinions, not a measurement — and it is unforgiving in practice, because doubling your amplifier power buys 3 dB and is barely a change, while sounding genuinely twice as loud takes ten times the power.

Your Meters Disagree Because They Measure Different Things

RMS stands for root mean square, a way of pulling one useful number out of a waveform that swings above and below zero. Averaging the raw swing gives you zero, so you square the values, average those, then take the square root — the result follows the energy actually in the signal, which is why it corresponds reasonably well to how loud something sounds. For a steady sine wave it comes out at 0.707 of the peak, about 3 dB below.

Music has no such fixed relationship. The gap between a signal's peak and its RMS is its crest factor, and a snare hit has a far bigger one than a sustained organ chord, which is why two meters on one channel refuse to agree. A VU-style meter averages over roughly 300 milliseconds — deliberately slow, built to show what a level sounds like — while a digital peak meter catches the highest single sample and ignores how long it lasted. Neither is lying. Clipping is a peak event, so watch peaks; balance is a question about average energy, so judge it on a VU or RMS reading.

Timing is only one reason two meters disagree. The other is the weighting curve, because your hearing is more sensitive through the midrange than at either extreme — a separate matter from everything above, and not why the decibel exists. An SPL meter bends its measurement to match, and which curve it uses changes the number you read. A-weighting was built from how people hear at quiet levels and throws away most of the bottom end: it treats 1kHz as its zero, and by 20Hz it is discarding roughly 50 dB. C-weighting is far flatter, which makes it the honest choice for a loud room, because your hearing flattens out as levels rise too.

Reading the Numbers on Your Own Gear

• Check the letter before comparing two specs — dBu and dBV are not interchangeable, and adding 2.2 to a dBV figure converts it to dBu.
• Expect about 12 dB where consumer meets professional, not 14 — a −10 dBV source into a +4 dBu input lands that far down, and preamp gain cannot give back the noise floor.
• A dB with no letter is relative — a fader reading tells you how far you have moved from unity, not how loud anything is.
• Switch an SPL meter to C-weighting for a gig — A-weighting discounts 50Hz by roughly 30 dB, so a bass-heavy room reads lower than it feels.
• Price an amplifier upgrade in decibels — twice the watts is 3 dB, four times is 6 dB, and ten times the watts is where most people would say twice as loud.
• Never read a dBFS number as a voltage — a meter at −18 dBFS tells you nothing until you know what analogue level your converter calls full scale.