in Tempest, Uncategorised

Tempest Recipes: Snare Drums

I’ll be honest – I’ve been fretting over this recipe for quite a while. I did a lot of reading and experimentation with snare synthesis and I had prepared a lot of background text on different snare types and how they might be approximated blah blah blah.

But then I came to a stark realization:  You probably don’t want to hear all this – it’s all on Google anyway – but more than that, although snares can be highly individual, the core snare sound is pretty standard and can be synthesized with ease. There’s no big secret. Once you know how to get the basic patch, there is a rich vein of snare variants to be mined from it.

In many modern music genres the snare tends to hit on the upbeat (beats 2 and 4) and serves to give a fixed focus to the entire rhythm. While other drum elements fly around, it is the snare that reinforces the sense of tempo. Thus the snare is often be the loudest instrument in the kit. It is an important reference for our ears. Changing the pitch of the snare can completely change the feel of a beat. To me, short and snappy snares make a rhythm sound more ‘urgent’, but this also depends on their placement.

Snare ‘ghost’ rolls are those barely-perceptible snare-hits in between that can add a lot of interest and variation to a beat without dominating. On the Tempest, the roll feature is perfect for real-time ghost rolls on velocity-sensitive snares.

Moving-on, for patching references we must define a standard. Our archetypal snare is a complex instrument. The stick hits an enclosed drum head, giving the initial impact transient which is then consumed by the rattling of the snares, producing a rich, noisy tail. But there is also some vibration from the second head which adds more subtle elements. Don’t forget that, these days, post-processing makes all the difference.  A snare without reverb rarely works.

 

Many standard snares, when viewed through a spectrum analyzer, will show a pronounced ‘body’ around 120-200Hz.

Here’s our representative example: Acoustic Snare

…and it’s spectrum:

Although this looks to be low-cut filtered, we can clearly make out a fundamental that peaks at around 200Hz. The rest of the sound is mainly the complex harmonics of the rattling snares, but there s a ‘body’ in there too.

To bring home the generic nature of this sound, here are the spectra for some well-known classic electronic snares:

TR-808:

TR-909:

LinnDrum:

..and finally , my personal favourite, the broody TR-505 snare:

Pictures speak a thousand words. We see the distinctive transient on the 200Hz region, with something of a frequency dip at around 1000Hz, and a tangle of high-frequencies. We can hear differences between these snares – but we recognize them as snares because they are basically the same sound done in a slightly different way.

Up to now we have touched upon the fundamental, but we haven’t discussed harmonics. This might be a good time to brush-up. A sine wave is harmonically ‘pure’ because it consists of only the fundamental frequency. Look at a sine through your spectrum analyzer and compare for yourself with the Acoustic snare above. There are complex harmonic relationships at play here, and they will vary widely (who wants 2 snares to sound the same?). We can’t fool ourselves into thinking that synthesizing these faithfully will be easy.

The engineers at Roland, Obeheim, Linn et al. encountered the same complexity and the proof is in the obviously synthetic nature of the snares on the classic drum boxes. The harmonic relationships are far beyond what can be modeled precisely. However, by following a standard recipe and USING OUR EARS it’s possible, with a bit of patience, to get close to any of the above.

Understanding harmonics is critical in digital FM synthesis which, BTW, is a rich playground for programming precision drums. Tempest’s FM implementation is too rudimentary for this purpose, but it is still useful.

By listening closely to our snares, some general points emerge that we must keep in mind:

  • The Impact of the snare comes from the initial hit from the stick and the resonating drum heads.
  • The attack transients are short, punchy, and louder than the other elements.
  • The envelope should start immediately loud for milliseconds and then quickly decays, leaving just the snares in the tail.

From this a simple but very effective snare can be synthesized with only one oscillator and a noise source. This is a solid foundation, thus your emphasis is getting the basics down so you can add and twist additional elements. Your goal should be the ability to dig-out your own snares from scratch in seconds, on-the-fly, mid-performance. Keep this goal in mind, and perhaps have a simple pattern playing to help guide your ears as you tweak.

  • We’ll start with a Triangle wave in Osc2 with a mix of 0/100. The reason we don’t want Osc1 to sound is because it is the Filter FM modulator. It may be needed later.
  • I usually choose to leave Osc wave reset off (because we want a little chaos) and keyfollow-on, so that we can play the snare at different pitches in 16 Tunings mode. Great for ‘found sounds’.
  • Using the signal analyser, pitch it down to the 200Hz region (F2). Listen out and you’ll know when it’s right. It’s usually better to do this while there’s a basic pattern running so you can tune the snare appropriately to the other drum sounds/instruments.
  • Apply a little pitch envelope – just enough to give a little extra interest without a noticeable pitchdrop. For me, a deacy of 40 and pitch amount 10 sounds good.
  • Ok now, let’s bring-in the ‘snares’. The Resonant 4K noise sample is ideal for a start because it can be pitched. The digital samples are by default much louder than the Analogue Oscs, so turn it way down to about 60 whilest listening for the correct balance. We should be in familiar territory now.
  • I prefer to use the 2-pole filter here to prevent resonant squeaks – but this might be just what you want. With resonance at 0, a Cutoff value of 40, slowly increase the Filter envelope amount and keep going. This should be sounding more convincing, if a little weedy. My approach is to leave the Amp envelope decay long and use the filters to shape the whole sound.

But now we run into a wee ‘limitation’ of the Tempest: The analogue Osc’s amplitude is hard-wired to the main Amp envelope, and you can only mix between them, not turn them down relative to the digital Oscs. This means that the amplitude of our digital noise – an essential element of the sound – will be controled by the same envelope. This won’t do because we want independent control over the amp envelopes of the skin (analogue oscs) and the snares (digital noise). We want the noise to have a smoother attack and a longer decay than the attack transient.

There are several approaches to this, but the simplest and most effective I’ve found is to use the Highpass filter as a surrogate Amp control. Noise is present in all frequencies, however the attack transient is, as we know, in the 200Hz region. So if we attach an aux envelpe to the highpass, we can close the sound-off  independently of the Amp envelope – which we keep open to allow the snares to ‘rattle’. This might go some way towards explaining for the 1000Hz frequency dip we see in the spectra above.

  • Attaching aux envelope 1 to Highpass, set an amount to about 90 so we can clearly hear the effect and play with the attack and decay of the envelope. The attack will need to be reasonably long – again use your ears to find the sweet-spot where the transient and noise transition is smoothest. At the same time, tweak the Lowpass cutoff, envelope and maybe add a little resonance to bring out those harmonics. There is a lot of tweaking here but the results are a-plenty.
  • Finally it’s necessary to adjust the Amp envelope – probably you will need to reduce the decay, and listening-out because this throws-up a surprising variety. Also, shorter sounds can reduce voice-stealing on the Tempest, so it’s wise to keep your sounds tidy.

By this stage you’ve probably gone off on numerous tangents, many of which may have ended-up as disappointments, however the gold is there if you keep digging.

Here are some more ideas to help make the snare your own:

  • Remember we were keeping Osc 1 for the FM? Oh yes – now we can try this out. I like to start with a pulse wave. Keeping Osc mix at 0/100 means that we don’t hear Osc 1 but it still modulates the filter. A rich ground for experimentation and lots of electronica surprises for the adventurous – especially if you modulate the crap out of Osc1 pitch. 🙂
  • With an 8-slot mod matrix per sound, there’s no excuse for not making your snares as dynamic as possible. My advice is to turn all envelope amounts down and turn their velocity amounts up. You may need to bash the pads harder, but it will almost ensure that no two hits will sound exactly the same, as is reality. A stronger hit means the sound will sustain longer – this is easy to model – use Modpaths to map velocity to Envelopes decay. A tiny bit of random mod mapped to fine pitch is also worth trying.
  • Also, don’t forget the ghost-rolls. With the roll feature-on, it’s simple a matter of gently applying variable pressure to a pad for endless snare fills.
  • Although I was reserving Osc1 for FM duties, it can also be brought into the fray. There is a story of magical frequencies in the TR-909 Snare – maybe you could try using Triangles at 185 Hz and 330 Hz 😉
  • Use noise as a modulation source! It’s available in mod paths and aloows you to turn anything to noise. Map it’s modpath amount to an envelope or fast LFO and see what happens. It could easily replace the noise samples if you map it to Osc frequency.
  • Use the LFO’s to make some chaos – especially on Osc frequencies, and especially Osc1 when used as FM modulator. Not to mention the filters…
  • You might want to change Pitch envelope destination to just Osc2 – but experiment – especially when using Osc1 as FM source, and when using the Pitched noise sample. Nobody says you can’t use a separate pitch envelope for each element.
  • Bring-in some extra noise – try the various colours, shaping their volume with envelopes. By default the Lowpass filter probably is squeezing out too many high-frequencies. You can open the filter/env some more or allow some of the noise to bypass the filter. Caution though – bypassing the filter affects both digital oscs at once, so use only a tiny amount of post-filter mix and rememer that this is controlled by the Amp envelope.
  • Control the extra noise with the other Aux envelope. Let it fade-in a litte after the transient and build-up to the ‘whoosh’.
  • Add some amp feedback – just enough to give extra ‘oomph’ in the attack transient without obvious distortion. Really adds a kick.
  • There is no shame at all in using the digital snare samples – in fact it’s encouraged – but layer them with the analogue Oscs for maximum profit.

Post-Processing is obviously important. Dry snares are rare – no wonder as these sound respond beautifully to a pinch of reverb. Maybe a dash of pitch-shifter and a little compression and you’re in business.