
A decibel (dB) is a logarithmic unit that expresses the ratio between two values — most often two levels of power or amplitude — rather than a fixed amount of anything. That single idea, that a decibel is a ratio on a logarithmic scale, explains almost everything that confuses people about it: why turning something up by 10 dB multiplies its power tenfold yet only sounds roughly "twice as loud," and why the same "dB" can mean completely different things depending on what it's measured against.
Start with what a decibel is not. A meter is always the same length; a kilogram is always the same mass. A decibel is different — on its own it doesn't name a quantity at all. It tells you how one value compares to another. "+3 dB" means one thing is about twice the power of another; "-6 dB" means one thing is about a quarter the power of another. Until you know what the "other" is — the reference — a lone decibel figure is only half a statement.
That's why audio decibels almost always come with a suffix: dBFS, dB SPL, dBu, dBTP. The letters after "dB" name the reference the ratio is measured against. Strip the suffix and "-14 dB" could describe a digital level, an acoustic loudness, or an analog voltage — three unrelated things. The suffix is what turns a bare ratio into a real measurement.
A decibel is logarithmic because each equal step represents an equal multiplication, not an equal addition. On a linear scale, going from 1 to 2 and from 100 to 101 are both "+1." On the decibel scale, every +10 dB multiplies power by ten, every +20 dB multiplies it by a hundred, and every +3 dB roughly doubles it. The gaps between the numbers stay the same while the underlying values grow explosively.

The reason audio adopted this scale is that human hearing works the same way. Your ears don't perceive sound in equal linear steps — they respond proportionally. A sound with ten times the power isn't perceived as ten times louder; it's perceived as roughly twice as loud. Doubling power (+3 dB) is only a small, perceptible nudge, while a change that feels like "twice as loud" takes about +10 dB, a tenfold jump in actual power. A logarithmic unit like the decibel compresses that enormous range — from the faintest audible whisper to a jet engine, a ratio of trillions to one — into a compact, human-readable span of numbers. Linear units simply can't hold that range without becoming unwieldy.
This is the crux of the "twice as loud" confusion. +10 dB is ten times the power but only about double the perceived loudness. Both statements are true at once, because "power" is physics and "loudness" is perception, and the decibel deliberately sits between them. Keep those two apart and most dB confusion dissolves.
Because a decibel is a ratio, audio has several decibel scales, each pinned to a different reference. They share the "dB" name and the logarithmic behavior, but they measure different things and their numbers are not interchangeable. Four come up constantly:
| Unit | Reference (0 point) | What it measures | Where you see it |
|---|---|---|---|
| dBFS | 0 = digital full scale (the maximum) | Digital sample level | DAWs, meters, exported files |
| dBTP | 0 = full scale, reconstructed | True (inter-sample) peak level | Mastering limiters, delivery specs |
| dB SPL | 0 ≈ threshold of hearing | Acoustic loudness in the air | Room levels, speakers, hearing safety |
| dBu / dBV | Fixed analog voltages | Analog signal voltage | Mixers, outboard gear, interfaces |
dBFS (decibels relative to full scale) is the one you meet most in software. Its zero sits at the top: 0 dBFS is the loudest a digital signal can go, and everything below it is negative. A peak at -6 dBFS is 6 dB below the ceiling. Push past 0 dBFS and the numbers run out, so the waveform flattens — that's audio clipping, and the space you leave below 0 is your headroom.
dBTP (decibels true peak) is a close cousin of dBFS that predicts a subtler problem. A digital file stores discrete sample points, but playback reconstructs a smooth wave between them, and that reconstructed wave can overshoot higher than any stored sample. A true-peak meter estimates those overshoots, which is why finished masters target a ceiling around -1 dBTP rather than 0 — leaving a decibel of margin for peaks a normal sample meter never shows.
dB SPL (sound pressure level) measures loudness in the actual air, not in a file. Its zero is roughly the quietest sound a healthy ear can detect, so the numbers run positive: a quiet room around 30 dB SPL, conversation near 60, a concert past 100. This is the decibel most people picture, and it's why prolonged exposure above about 85 dB SPL is a hearing-safety concern.
dBu and dBV live in the analog world, referencing fixed voltages rather than a digital ceiling or air pressure. You meet them at the boundary between hardware and software — mixer outputs, audio-interface specs, outboard gear — where a signal's level is a real voltage on a cable. Most browser-based work never touches them, but they explain why "line level" is a specific number, not a vibe.
The takeaway isn't to memorize the table. It's to notice the suffix. When someone says "keep it under -1 dB," the useful question is dB relative to what — full scale, true peak, or something acoustic — because the same number means different things on each scale.
Once you see the decibel as a ratio, it stops being one topic and starts being the unit underneath most of the others. Loudness targets like -14 LUFS are decibel-based measurements of perceived energy over time. Dynamic range is a dB figure: the gap between a track's quietest and loudest parts. Headroom is a dB distance below the ceiling. Clipping is what happens when a signal exceeds 0 dBFS. Compression works by pulling loud parts down by some number of decibels once they cross a threshold. Even the way we talk about audio frequencies — a bass boost of "+3 dB at 80 Hz" — uses decibels to describe how much a band is raised or cut. Learn the unit once and every one of those concepts gets easier.
You can work with all of these directly in the browser. On vocalcut.com, every tool runs entirely on your device — your audio never leaves your machine, and there's no account or upload step (the processing engines download to you, not your files up to a server). When you record audio in your browser, the input meter reads in dBFS, so you can set a level that peaks well below 0 and keeps clean headroom from the start. The volume normalizer measures a file and applies gain in decibels to hit a loudness target without reshaping the sound, and the in-browser mastering chain shows peak and loudness in dB while you set a true-peak ceiling on export. In each case, the decibel is the language the meters speak.
Decibels look intimidating because one word covers so many scales, but the core is small: it's a logarithmic ratio, each 10 dB is a tenfold change in power and about a doubling in perceived loudness, and the suffix tells you the reference. Hold onto those three facts and the meters across every audio tool suddenly read as one consistent language.
Why does +10 dB sound twice as loud but mean 10 times the power? Because power and loudness are different things. A +10 dB change multiplies the actual signal power by ten, but human hearing is logarithmic and perceives that tenfold jump as only about twice as loud. The decibel is defined around the physical ratio (10x power per 10 dB), while our ears compress that ratio down to a roughly doubled sensation. Both numbers describe the same change from different angles — one physical, one perceptual.
What's the difference between dBFS and dB SPL? dBFS measures a digital signal's level relative to full scale, where 0 dBFS is the maximum a file can hold and everything below is negative. dB SPL measures actual acoustic loudness in the air, where 0 is roughly the threshold of hearing and the numbers run positive. They share the "dB" logarithmic behavior but reference completely different things, so a dBFS value and a dB SPL value can't be compared directly.
Is 0 dB silence? Not usually. Zero decibels just means the measured value equals its reference, so what "0 dB" means depends on the scale. On dBFS, 0 is the loudest a digital signal can go, and silence sits far below it (very negative). On dB SPL, 0 is roughly the quietest audible sound — near silence, but a specific reference, not true nothing. A bare "0 dB" is meaningless until you know the reference.
Why is 0 dBFS the maximum instead of the minimum? Because dBFS is measured down from the digital ceiling. Full scale is the largest number the samples can represent, so it's defined as the 0 point, and any real signal sits below it at a negative value. It's the opposite orientation to dB SPL, where 0 is the quiet floor and levels climb positive. This is why digital meters show negative numbers and why "louder" means "closer to 0" in a DAW.