Compression Basics – Compressing live percussion

Live percussion recordings tend to have a large dynamic range, and hence are a great vehicle to use to learn the basics of compression.


Compression seems an appropriate topic for my first ‘howto’ article, given that it’s the effect that I’ve learned by far the most about over the last 6 months.

I tend to use a lot of recordings of live sounds in my tracks, particularly for percussion.  Live percussion recordings tend to have a large dynamic range, and hence are a great vehicle to use to learn the basics of compression.  Because the dynamic range is so large, it requires either a single pass of a compressor with very aggressive settings, or better, successive applications of more gentle compression and limiting (as I’ll show here).

Applying compression is often a difficult technique to learn, because the differences imparted by mild compression (e.g. with low ratio or high threshold, for example on a master bus) can be difficult to recognize unless you really know what to listen for.  However, when reducing the dynamic range more dramatically (as is usually required on live percussion samples) it’s much easier to hear the effects of the compression

The example sound I’ll use is a recording of a steel drink can knocked against a hard table (recorded using a Rode NT3 condenser microphone). I thought it was an interesting sound and wanted to save it in my sample library so I could potentially use it in a track at some point in the future.

One important point here is that you’ll need to use good headphones or monitor speakers to properly hear the difference between the audio samples below.  It will likely be difficult to hear the differences properly on laptop, tablet, or phone speakers.

The raw sample sounds like this…

and has the following waveform…


…straight away you can hear (and see) the big difference in level between the initial transient sound of hitting the table, and the ‘tail’ sound of the can ringing (starting from about 0.015 seconds).  If you tried to use this sound in a track as-is, you’d have to keep the level of it fairly low to prevent the loud transient from clipping, and then the nice, harmonic ring of the can would probably be completely lost under other sound layers.

For these types of samples, I usually firstly apply some limiting to reduce the level of the transient peaks (using Waves L1).  In these cases I often find looking at the waveform more closely helps to give you an idea of where to initially set the threshold of the limiter…


In this case, I ideally want to trim the two most prominent peaks at the level marked by the red lines.  These show a 16bit integer value of 18,000, which equates to roughly -5.2dB (note that waveform axis is marked as 16 bit, although the sample itself is 24 bit).  Auditioning L1 on the sample, I was actually able to limit down to -8.2dB threshold without adversely affecting the sound.  Also, because we’re just limiting peaks in this case which rise and fall very quickly, I’m using a very short release value.  Ultimately I used the following settings in L1…


… and it resulted in the following sound and waveform changes…


Zooming in on the waveform again, what I want to achieve is to further reduce the difference between the peaks and the ring of the sound… visually, to try and ‘pull’ the peaks more towards the red lines.  Using limiting again would be too harsh, and would probably remove the dynamic and ‘impact’ out of the sound… hence I use a compressor (Waves C1).  Again, using the red lines as a guide for the initial threshold setting, these are at 16 bit value 6000, which equates to approx -14.7dB.  I probably want to reduce the level of these peaks by about a 1/2 or a bit more above the threshold, so I would guess at a compression ratio of around 2:1 to 2.5:1


As with the limiter, I’m trying to just ‘pull down’ transient peaks here whose wavelength is very short (a handful of 44.1Khz samples), so I use short attack and release settings in C1.  From previous testing, I’ve found the quickest attack and release settings you can use in C1 without it introducing undesirable artifacts (‘clicking’ sounds as the compressor engages) are about 0.04 and 30ms respectively.

Ultimately I used a bit higher threshold than the -14.7dB estimated.  The reason for this that C1 has a fairly soft ‘knee’ (i.e. it starts introducing compression gently as the sounds approaches the threshold level).  I looked at the waveform to get a rough idea of the initial threshold and ratio settings to use, but these need to be auditioned and finalized by ear.  I settled on the below settings, which gave a nice balance of still having some dynamic and ‘impact’ but allowing the ‘ring’ part of the sound to be closer in level to the transient (it showed around 3dB of gain reduction on the meter in C1).  The final step was to add 1.9dB of makeup gain, which audibly level-matched the compressed sound with the original.


If I was using this sample immediately in a track I probably would have gone for slightly more aggressive settings (less threshold or more ratio), but given it’s to be put in my sample library, I erred towards more conservative settings to make the sound more generally useable.  The resulting sound and waveform are below…


At this point I normalized the level of the sample up to -3dB.  The final step I usually take with these kind of samples is to do one more application of L1, just to trim the highest peaks, but without changing the sound of the sample.  This is just to try and reduce the transients as much as possible, which makes the sound easier to mix into a track without master bus clipping (I’ll discuss in more detail in a later post).  Looking at the waveform again to give me a guide for the initial settings, I want to try and contain the peaks to within the red lines (16 bit value of 20,000 ~= -4dB).  I used a threshold of -4dB in L1…



…which trimmed the peaks, but without noticeably changing the sound.

With these types of percussive samples, the last step I take is usually to use a gate to fade out the tail of the sample.  The appropriate gate settings are best judged by ear, and I settled on those below, which I thought gave a nice balance between allowing some of the ‘ring’ of the can to sustain, but also fading out the hiss of the noise floor (and I used the Waves C1 Gate for this)…


If you compare the initial raw sample against the final one below, the final one has a lot more evenness between the transient and the ‘ringing’ tail of the sound… the whole of the sound can be heard more clearly, and this will make it far easier to mix into a track along with other instruments.  It has a ‘stronger’ sound than the raw sample, but peaks at a lower level.

For readers who are a bit unsure of the appropriate applications of compression and what settings to use (as I once was), I’d encourage you to try the above steps with your own live percussion samples.  For me it was a really good way to practically understand the effects of compression, and to be able to clearly hear the results.

4 thoughts on “Compression Basics – Compressing live percussion”

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s