To compress or not to compress?

Now we have something on tape, the first thing to do is listen back, carefully.  Be critical.  Are you happy with the source recording?  Is there too much or too little room?  Have you picked up too much pick?  Does the performance sound flat?  These are all good questions to ask yourself before moving on.

The ideal is to do, not too much.  Old-time is very much about capturing the performance and enhancing it in a sensitive way.  Don’t be afraid of doing several takes, changing the setup each time, see what works.  The better the source recording the less you do in post.

If you’ve found a sweet sounding space, if you’ve got a good signal going into the recorder and you’re happy with the result, then we’re more than halfway there. 

I thought the first thing we’d take a look at is compression…  This is a complex topic with many ways we could go.  The following is a look at basic compression.  This will help smooth out any peaks you may have which would allow a small increase in the overall level.

I guess this is a good time to say what I use for post production.  I use two software packages: Sony Sound Forge Pro v10 and CakeWalk Sonar 8.5.  For all my actual editing I use SF and for any multi track tracking or other work requiring the mixing of audio sources I use Sonar.  I also use a Contour Shuttle Pro hardware controller/mouse.  A most valuable bit of kit when it comes to fine editing work.

There are many WAV editors out there, from free-bees such as Audacity to high end ones such as SADiE.  Whichever you settle on using, the important thing is to make sure you have a compressor with the following controls: threshold, ratio, attack, release, make up and input/output gain.

There are number of budget wav editors out there such as Sony’s Audio Studio which is a cut down version of SF Pro.  Others such as Goldwave I have found to be less intuitive.  The compressor in Audacity is, OK, but it seems to work in percentages rather than dBs which I find odd.  It is possible though to buy a third party compressor and import it into the effects list.

Compression is a much missed understood and miss used post production tool; mainly due to the difficulty in hearing what it’s actually doing and the fact there are no fixed rules or guidelines.  The temptation is to crank it up so you can hear a difference.  But all this will do is make your recording sound hard sounding.

Let’s first of all clear up what a compressor does.  Its basic function is the levelling of overly loud sounds, ie those sounds that are naturally high compared to the average level of the recording.  For example, we have a recording where the average level is say around -10dB.  There are a few spikes reaching say -1dB.  So the most we can raise the level of the whole recording is only by 1Db before reaching 0dB.  The job of the compressor is to squash the high peaks and bring their level down nearer to the average.  So if the peaks are compressed so they only read -4dB, this would allow us to raise the overall level of the audio by 4dB.  This has the effect of bringing up the quieter sections by 4dB and giving the illusion of greater volume.

A brief overview of the controls:

• Threshold – this is the point at which the compressor starts to compress.eg, -8dB any peaks after 8dB are reduced.

• Ratio – This is how many dB the signal is reduced by. eg for a ratio of 2:1, every dB past the threshold, the signal is reduced by 2dB.

• Attack – measured in milliseconds (ms), this is how quickly the compressor kicks in.

• Release – Measured in milliseconds (ms), this is how quickly the compressor releases the reduction.

• Make Up – allows you to raise the overall level of the audio.

• Input/Output gain – allows you to change the actual output or input level coming in or going out of the compressor.

This is a little crude, but more or less what happens.  There may be other controls such as: look ahead, smooth saturation, auto gain and so on.  Each type of software compressor will have different characteristics: this means they will not necessarily give the same result when using the same settings.

So, what would we need to do to our recording?  We’re not looking to hammer it hard up to 0 like many modern pop and rock bands would do.  We’re looking to preserve some of the dynamic properties of the performance but at the same time levelling off some of the extreme peaks.  This then may have the benefit of allowing us to raise the overall level by a few dB.

Before we start, a note on “noise floor”.  The noisier the background the higher the noise floor which means when we come to raise the level of the recording we are also raising the level of the noise floor.  For example, in the sample recording available at: http://www.clearvoicemedia.co.uk/ch-samples.html the noise floor of the source audio is about -63dB or so.  I was actually surprised at how good this was.  So lifting the level by 3 or 4dB won’t make much difference to the overall background noise.  But, if the noise floor is only -42dB, then lifting it by 4dB may make an audible difference.

Here’s an excellent article on dB from our friends at Wikipedia:

https://en.wikipedia.org/wiki/Decibel

Where do we start?  Well, the first thing to do is listen through the tune and make a note of the high and lowest points.  For example, you may have a low point of -9dB and a peak of -1dB.  This gives us a dynamic range of around 8 to 9dB.

Once the dynamic range of the tune has been established then the next thing to do is dive in…

To give you an idea of crunching the peaks hard, try the following:

• A threshold of -25

• A ratio of 7:1

• An attack of 30ms

• A release of 70ms

• Adjust the make up according to the output of the compressor

See what happens…

But, as we’re not looking to crunch the peaks to hard a good starting place would be the following:

• Threshold of -8 to -12dB

• Ratio of between 1.2:1 to 1.7:1

• An attack of between 0.25 and 30ms

• And a release of between 100 and 500ms

• Make up the gain according to the output of the compressor

• Or adjust the input/output gain accordingly

The above is a rough guide only.  You really have to play around and listen carefully.  Start with some fixed points, ie fix the threshold at -12dB, fix the Ratio at 1.5:1, the attack at 3ms and the release at 350ms.  Listen carefully, make comparasons between these settings and the source by using the bypass control.

Then play around with the ratio or attack, see what happens.  Don’t forget to use the bypass to make an A-B comparison between your source and the compressor.  Then do the same for each of the other settings.

The aim is to smooth out the performance, ie to lower the maximum peaks so creating a smoother balance between the quiet and loud passages or notes.  So if your audio peaks at 2.4dB, if working within the above settings it dropped to -4.8dB then you’re on the right path.  Depending on your dynamic range, I reckon if you are lifting the audio by 2.5 to 4dB then that’s probably about right.  We’re only looking to smooth out some of the more excitable moments of the performance.

It’s definitely worth playing round with the attack.  If you’re using picks and they sound a little obvious in the source recording.  It’s possible to reduce the clatter by using a fast attack, say 0.5ms.  Basically the compressor is kicking in at the start really quick and reducing the initial transient.  Be careful though as you can easily flatten the rhythm of the tune.

Just for general interest, ratios roughly fall into the following categories:

• 1.01:1 – 2.0:1 – soft

• 2.01:1 – 5.00:1, medium/artistic

• 5.01:1 upwards, Hard/limiting

Some compressors only allow the 1 decimal place so just round up to that.

It will take time to start to understand what you’re hearing.  To be frank, compression is something I’ve struggled with for some time.  I’m still not convinced I always hear or understand what’s going on.  Most of the time it’s more of a gut feeling.  It’s just trial, error and more of the same.

You can find some samples of compressed audio at:

http://www.clearvoicemedia.co.uk/ch-samples.html

They may help, they may not.  Please feel free to download the source audio to try out compressing it yourself.

I will come back to this topic later.  There’s much more that could be said, but it’s worth playing round with the basics first.

Now I’m afraid this is where I have to leave you to play around and exercise those ears!  My next subject will be basic EQ…

Recording clawhammer Banjo