## Monday, August 6, 2012

### SNR 101 - The Decibel

We talk a lot in this industry about decibels (dB) but they can be one of the most difficult concepts to grasp. The short answer is that a decibel is one tenth of a Bel. Since that gets us exactly nowhere let's get into what it really is. It's a way to compare values that might differ widely by using ratios along a logarithmic scale.  For instance you can say the difference between one watt and 10,000 watts is 40dB. That's because a logarithmic scale is labeled 1, 10, 100, 1000 instead of 1, 2, 3, 4.

As usual the purpose of this blog isn't to write the last word on the subject, just to help people get a handle on the concept. If you really want to get deep into it there are a ton of resources and calculators out there on the web. Bring your reading glasses and number two pencils it gets deep fast.

So what's useful about decibels besides not having to write out five digit labels on tiny knobs on sound equipment. It turns out that the human ear and brain interpret sound on a logarithmic scale. That's why you can hear a bee buzz and not collapse in pain at a thunder clap. The amount of acoustic energy is vastly different but they're not that far apart on a decibel scale.

So here's the first useful tidbit. If you raise a signal 3dB you've doubled the amount of energy. Stop there though because you haven't doubled the volume. Because sound propagating through space behaves according to the inverse square law (put the pencils away I'll try and make it simple) it takes an increase of 6dB before you hear something as twice as loud.

OK, I tried to make it simple and it wound up running to four paragraphs and I may have plagiarized Wikipedia extensively so let's just forget it and get to how all these numbers work for you on stage and in the studio.

Let's start with the stage. If you're running a 10,000 watt system and you're going flat out and tripping breakers, dropping your level at the main output by one dB will shave over 2000 watts off your energy budget. There's a lot more to it than that but average levels, compression and efficiency aside, you can save quite a bit of power with a small adjustment. Drop your main outs by three dB and you turned your 10,000 watt system into a 5000 watt one, but you haven't cut the volume in half. So if you cut your mains down to save some electricity you can then use a touch more compression to sneak the average level back up and have things appear nearly as loud.

In the studio it's much the same but the enemy is the hard limit of the digital clip. At a gig you're looking at dBu, that's decibels relative to unity gain, in the studio you're usually looking at dBfs which is decibels relative to full scale, or when you run out of bits and it clips. In trying to make things louder on CDs, MP3s and the radio studios use more and more compression to get that apparent loudness up there.

There's no bigger sound system in the world of recorded audio. Zero dBfs levels the playing field. So the difference between two records that both peak just a hair below that is how much they compress the signal. A record that uses almost none could have a dynamic range of a couple dozen decibels, between a quiet piano line and a full band blasting through a bridge. That's pretty interesting to listen to but can seem a lot less exciting when compared to a pop record that's compressed so there's only 6dB total dynamic range. A whisper is literally half as loud as the full band blasting away.  It sounds like the loudness wars are starting to come to an end, and people are willing to listen to stuff that's more interesting instead of just loud, but it's still really important to look at how many decibels of range your material covers. You don't want that touching piano riff that's all by itself to get drowned out by wind noise if someone is listening in a car.

Well Brethren of the Knob and Fader, that wound up being pretty long for a brief explanation of the decibel but I think it gives you an idea of what a complex thing it is. This is where you go off and dig a little deeper so you can show up to your next gig a little better informed. Get going!

1. To add just another bit about the way we hear and how it relates to loudness in recordings / stage; it's useful to know that different frequencies, when heard at the same dB level have different perceived loudnesses. The implications of this are vast, but one helpful application is rolloff. If your mix is peaking (according to the meters), but still doesn't sound all that loud, it might be because you are pumping a bunch of low end through the system, but aren't aware of it because you can't hear it. Attenuating / getting rid of any unnecessary frequencies is the best way to make your mix "louder". And you don't have to sacrifice dynamic range to do it!

1. Teriffic comment Fred. I've talked a lot about apparent level due to compression or lack thereof and separately about clearing out the bottom end. Thanks for bringing the two ideas together on the same page.

It also makes me want to get my thoughts together on the topic of the Fletcher-Munson curve and put up a post on that.

2. Excellent information provided.I also want to share some information on sound level.
100 NORMAL AVERAGE CAR OR HOUSE STEREO AT MAXIMUM VOLUME
117-123 HOME STEREO SYSTEM, A VERY LOUD AND POWERFUL 200-2000 WATTS
120-130 FRONT ROW AT A ROCK CONCERT- UP TO 200 REFRIGERATOR SIZE SPEAKERS AND 50000-300000 WATTS OF CLEAN, FULL FREQUENCY SOUND
130 (N)MARCHING BAND-OVERALL LEVEL AT A DISTANCE, 100-200 MEMBERS
142 (P) INSIDE A CAR WITH TWO PRO 18 INCH WOOFERS AND 300 WATTS EACH
decibel levels

1. Like anything there are some qualifications to all that. What is harmful to the human ear depends largely on frequency. A system with no subs can be painfully loud at 95dB while one with full bass response can go above 100dB and you'll still be able to talk to the guy next to you.

3. If you're not familiar with the NPR show Radiolab, you should be. Regardless, here is a fun short about sound and decibels.