dB-scale and its "zero"

Hello there!
When I switch the scale on a sound file from “linear” to “dB”, I noticed something. The scale uses -60 dB as its zero, why is that?

I am doing my High school diploma project and this would be useful to know. I will insert a picture, please take a look at it so you understand what I mean

As dB is a logarithmic scale, the zero point for it is at -inf, and it’s very difficult to build a thing with evenly spaced ticks which goes all the way to negative infinity. Impossible, really.

For audio purposes, -60dB (or 0.001 amplitude) typically is where useful information ends, anything quieter than it tends to be just noise. Having it stop there is just convention though; you can change it in Preferences → Interface → Meter dB range.

Also, for Audacity 3.3, we’re adding a view which uses the linear scale but displays corresponding dB values:

Also, note here that it is a negative number. 0 is a point at which, once the sound is louder, the peak (or trough) will start clipping.

-10 means you could, ostensibly, increase the volume by 10 more decibels before you will start to hear distortion.

-60 is VERY quiet. But -INF would be what you’re thinking of as “Zero” in this case, (or absolute silence.)

In this scale, zero is actually MAX volume (without clipping or distortion.)

1 Like

Also, decibels are relative so you need a reference. With digital audio 0dBFS (0dB Full Scale) is the maximum you can “count to” with a given number of bits and digital dB levels are usually negative

16-bit signed integers can hold values between -32,768 and +32,767. If you have a digital file with those negative and positive peaks, it has 0dB peaks. You can’t count any higher without more bits.

Silence (-infinity dB) is a series of samples with a numerical value of zero. And there is a zero-crossing twice per cycle. A digital file may not have an exact-zero number, depending on where it’s sampled… The zero-crossing is usually in-between a positive and negative sample and the actual zero-crossing only occurs with the reconstructed analog output.

24-bit audio files have bigger numbers and 8-bit files have smaller numbers but when you play the files everything is scaled automatically to match the bit-depth of the DAC and a 24-bit file is not “louder”.

With floating point audio, 0dB is a numerical value of 1.0 and for audio purposes there is essentially no upper or lower limit. Internally, Audacity uses 32-bit floating point so Audacity itself won’t clip. It will “show red” for potential clipping if it goes over 0dB.

For acoustic loudness (sound in the air) the 0dB SPL (Sound Pressure Level) reference is approximately the quietest sound that can be heard so dB SPL levels are positive.

There are also electrical dB levels (such as dBV).

Normally, none of these levels are calibrated but they are directly correlated. If you change the digital level (or the electrical level) by -10dB (more negative) the dB SPL level also drops by 10dB.

Decibels are logarithmic (or “proportional”) like our hearing - If you have a speaker playing at 1 Watt and you double it to 2 Watts, that makes a 3dB increase. A small but noticeable change on perceived loudness. If we have a speaker playing at 100W and we increase it by 1W, it’s also 1W “louder” but proportionally it’s a small unnoticeable change of 0.043dB.

1 Like

This topic was automatically closed after 30 days. New replies are no longer allowed.