Facebook social media iconTwitter social media iconInstagram social media iconSubmit to Reddit

An Introduction to Compression: Basic Compression - A free download from Audio Masterclass

Equipping Your Home Recording Studio - A free download from Audio Masterclass

An Introduction to Equalization - A free download from Audio Masterclass

Five12 Releases Numerology 3

A brief introduction to equalization for the home recording studio

Can curtains provide good soundproofing?

How to get a 'vintage sound' in your recordings

Universal Audio Twin Finity combined tube/transistor mic preamp

Your school grades you 0 to 100%. But what does a real-world client think of your work?

Spectre 1.5 Takes Audio Analysis to Brilliant New Heights

Choosing equipment? Here's how to make the right choice and not waste money.

Your mix sounds good in your car. But does it sound good in ANY car?

What is production? Part 4: Mixing

Headroom - What is it? Why do you need it?

Recording engineers commonly talk about headroom. But do you need it? How do you get it? What happens if you don't have enough?

The term makes an analogy with a truck going under a low bridge. Suppose the truck is 3 meters tall and the bridge is 4 meters high across its full width.

No problem, plenty of headroom and the truck can breeze through.

But what if the bridge is 3.02 meters high, just a mere two centimeters higher than the truck?

Well in theory all should be well. But what if the truck's tires were pumped a little bit higher than normal? What if it was empty and riding high on its suspension?

Clearly, there is the potential for a nasty accident.

So we can see in this situation that headroom is desirable. Headroom costs money, because it generally costs more to build a bridge with a higher opening. And trucks that are not so tall can't carry as much payload.

So what is headroom in sound engineering? Simply, it is this...

Headroom is the difference in decibels between the highest level a system can take without distortion, and the highest level the engineer expects to use.

Let's take a simple, uncomplicated, example...

You're making a live recording of acoustic instruments using two microphones plugged directly into a portable recorder, such as a DAT digital recorder. (I use that as an example, because they are all very similar. Disk and flash memory recorders are often very dissimilar, although the same concepts apply.)

The DAT recorder has input gain controls, and a meter that shows the level going to the tape. The maximum reading on the meter is 0 dB, and all the other calibrations are in minus numbers.

So if the maximum level shows as -10 dB, then there is 10 decibels between that and the peak level the recorder can handle without distortion.

So, let's say we have a live recording of acoustic instruments. And for some reason, you haven't been able to attend the rehearsal. You can guess pretty well where to place the mics, but you have no idea of what gain to set because you can't know for sure how loud the band will be in advance.

So what do you do? Answer... you set an input gain based on experience. With typical capacitor mics used as a main stereo pair on acoustic instruments, an input gain of 40 to 50 dB would be typical.

But you have to err on the side of caution. What if they are going to play really loud? If you set the input gain too high the loud parts will go over 0 dB and will be clipped and distorted. And there's no cure for that.

So to be on the safe side, you set the input gain quite low. This will mean that the signal-to-noise ratio of the recording isn't as good as it could have been. But at least you will be sure that there will be no clipping.

So the concert starts. Actually the band isn't all that loud. In fact, the signal level never gets higher than -20 dB. You might consider raising the input gain, but you don't know whether they will get louder later on.

But as it happens, they stay at the same level all the way through.

That 20 dB of potential level that you didn't use was your HEADROOM.

In retrospect, it might seem like a waste, and you could have achieved a recording with a 20 dB better signal-to-noise ratio.

But it was worth it to be very sure there would be no clipping.

Now let's take another situation. You're still recording to your DAT recorder so nothing else has changed.

But this time you are mixing a multitrack recording to stereo DAT.

Typically you will play the multitrack recording through a number of times while you balance the levels, EQs and everything.

Eventually there will come a point where you are happy with your mix and you are ready to commit to stereo DAT tape.

So you set the input gain of the DAT (line inputs this time, but the concept is the same) so that at no point during the track do the red lights come on.

But this time things are different. The level coming into the DAT is TOTALLY predictable. It's the same every time you play the multitrack.

So because of this, you don't need any headroom. You can record right up to the last meter segment before the red LED comes on. The result will be a recording with optimum signal-to-noise ratio.

Setting the right amount of headroom is something that the engineer needs to think about, depending on the predictability of the sound source.

If it is completely predictably, you don't need any headroom at all. If it is fairly predictable (like a recording where you have had chance to rehearse), then you need a moderate amount of headroom, perhaps 10 dB. If the source is unpredictable, then even 20 dB of headroom isn't too much.

Far better to be safe than sorry.

There is another way to look at headroom, and that is the headroom that is inbuilt into the equipment. It means the same thing, but the engineer doesn't have so much control. We'll look at that another day.

By David Mellor Wednesday July 11, 2012