Drum Science - Part 3

Welcome to this podcast series on the science of drums, where we're discussing drum acoustics, drum kit setup, and approaches to drum recording and production. My name is Professor Rob Toulson. I've studied the science of popular drums for over 20 years. And in this series, I'm repackaging some of that scientific theory on drums in a practical way so that it can be useful to all drummers and studio engineers. In this episode we'll look in detail at how you can optimise the drum setup in a recording studio, and capture exactly the sounds you're aiming for. Now there are a number of things to consider when recording drums, for example, choosing the right recording space from an acoustics and technical perspective, deciding on the recording setup, how many microphones you'll use and where they will be placed, And we should also consider the mixdown at the recording stage too, and ensure that the desired end result can be achieved with the captured recordings. Before we move on to discuss those things, we first need to discuss a technical topic that is more relevant to recording drums than to any other instrument, and that topic is phase coherency. The concept of phase can be quite a challenging one to understand. Predominantly because it's a purely mathematical term, which is regularly misused and misrepresented in the world of music production. In fact, when the term phase is used in a music studio, what is actually being discussed is usually either time delay or signal polarity. So first I'll explain a little background of these terms, and then discuss how they relate to music recording, and specifically to drum recording. Well, time delay is quite a simple term to understand. If we measure the time which an event occurs, then we can measure if there is a delay in time relative to some other event. So this could be a delay between a repeating audio signal, as in a guitar delay pedal effect, or a delay in the time that two different microphones capture the same sound event, because the two microphones are at different distances from the sound source. We also get time delays from sound reflections from walls and objects in the studio. This results in a microphone and even our ears hearing two sounds spaced apart in time. Our brains usually interpret this as a natural reverb or echo, but in reality it's the result of lots of different time delayed sounds blending together in a closed space. We also get time delays digitally and in software. For example, if a computer processes the same sound twice, but with a small time difference in the two different processes. You might have heard this if you've recorded yourself singing without a perfect monitoring system, which can result in a small time delay in the sound that you hear back into your headphones. Now time delays can also be described in pure mathematics as phase shifts. This complicates things somewhat unnecessarily for musicians, because phase shifts are mostly relative to frequency and pure sine waves, which are measured with angles in degrees or radians, rather than the measurement of time itself. In fact, a single time delay, say 5 milliseconds, causes each audio frequency to have a different amount of phase shift. So let's hear the effect of time delay on an audio signal, and then discuss how that relates to the mathematical term fades. Okay, here's a mono recording of some drums being played. I'm now going to copy the audio file onto another channel. The result is that we hear the drums louder, because we now have greater amplitude of the two identical signals mixed together. Now I'm going to offset one of the audio files by a 10 sample time delay, which means about 0. 2 milliseconds. Can you hear the difference between the original and with the time delay? So here's the original without delay, and here's with 10 samples delay. Now let's increase the delay time. We'll start from the original audio file and then add the sample delays. Here's the original file. Now we add 10 sample delay again. 20 samples. 40 sample delay. 60 samples. 100 samples. 200 samples. Can you hear that every time we change the delay time, the sound of the drums changes quite significantly. This is the effect that small time delays have on sound. It's an effect that we often refer to as comb filtering, and can also be called phase incoherency. So what's actually happening here? Well, I mentioned that a single time delay causes all frequencies to have a different amount of phase shift. And a single sine wave, or frequency, is totally cancelled out if it's mixed with a version of itself that has been shifted by 180 degrees. But because sine waves repeat after 360 degrees, a whole bunch of other frequencies are also cancelled out by the time delay. So when we add a time delay to a signal and mix it with the original, we cancel out a number of frequencies. If the time delay changes, then the frequencies that are cancelled change too. Essentially the result is like applying a drastic EQ filter to the audio, but it's a filter that we can only control by managing the time delays of our signals. We call this effect comb filtering, quite simply because if you plot a filter chart of the effect, or repeat the experiment I just demonstrated with white noise, you see frequencies being filtered in a manner that resembles the profile of a plastic hair comb. The phase cancellation occurs at regular intervals, so we see regular dips in the frequency spectrum of the effect. This effect can actually be used creatively. By modulating the time delay, we get the well known phaser effect that appears on the pedal boards of many guitarists. In the studio, comb filtering occurs most with reflected signals and with setups using multiple microphones. So there's no surprise that it affects drum recording considerably, since we often use 10 or more microphones to record a drum kit. Drums also suffer from comb filtering because the frequencies of each drum stay the same during the performance. When playing piano or guitar, or any melodic instrument, the performer moves around and performs different musical notes. So the effect of comb filtering is less easy to recognize. But with drums, if one of the frequency cancellation points is exactly at the fundamental or overtone frequency of one of our drums, then there can be a significant loss in power or definition for that drum. Let's hear an example of that. We're going to listen to a drum recording and focus on the snare drum, listening to both a close mic and an overhead mic. Here's the close mic of the snare, which sounds full and powerful. Here's the overhead recording of the kit, which also sounds good, though with much more roomed sound. Now, because the microphones are positioned at different distances from the snare, there is likely to be some frequency cancellation. And if we're unlucky, that frequency cancellation could be at one of the main frequencies of the snare drum. So let's first hear the close mic recording, then the overhead recording, then the two mixed together. Here's the close mic, and now the overhead, and here the two mixed together. If you have a good playback system, you should be able to hear that the snare sounds deep and powerful in each mic individually. But when they are mixed together, a powerful low frequency of the snare gets lost. So it appears that the fundamental frequency of the snare is being cancelled out or attenuated when we mix the two microphones together. And if I zoom in on the waveform, I can see that the most significant sinusoidal component of the snare drum is significantly out of phase between the two signals. So this confirms what we are hearing. But how can I resolve this issue with frequency cancellation in the snare drum? Well, there are usually three or four options. Firstly, we could move one of the microphones to change the amount of time delay. But this has the effect of changing the tonality of the recording and the balance of reverberation. For example, if we move the overheads further away, they may start to feel too distant and be too heavily weighted with room reflections. We could change one of the microphones. All microphones have their own inherent response characteristics, and you may find that a different mic on the snare has no issue with frequency cancellation at all. Thirdly, you could nudge the waveforms in post production and make them align perfectly. But with this approach, you need to be careful that by correcting one issue, you are not causing other issues related to different drums and performance timing. The simplest potential solution is to flip the polarity of one of the signals. And this is easy to do, either on the mixing desk or in post production. Flipping or switching the polarity means inverting the signal by making all positive amplitudes negative, and making all negative amplitudes positive. This is very easy to do in computer software. We just multiply every sample by a value of minus one. It's also really easy in an analog circuit, because most microphone signals are what we call balanced, which means they carry both positive and negative forms of the audio. So a simple electrical switch can flip positive to negative, and vice versa. This is why you'll regularly see a polarity switch on the channels of most mixing desks and outboard preamplifiers. Switching polarity is a rather blunt solution, because it doesn't actually give us any control of time delay or phase shift of frequencies. Which is the real issue that we are trying to correct. But in a surprising number of circumstances, polarity switching actually works and gives an improved result, and has little side effects to worry about. Sometimes the reason this works well is because microphones can be pointing in different directions. Such as a microphone pointing down on top of the snare, and another pointing upwards on the underside of the snare. So the captured signals may have opposite polarities in this instance too, and can benefit from being rectified. Let's go back now to our snare drum example and see if polarity switching helps in that scenario. In post reduction I can add a polarity switch and evaluate if the sound is more complete with or without the switch engaged. Let's hear the overhead and close snare mic together again. Now let's flip the polarity of the snare mic. And let's hear with the switch off. And the switch on. Switch off, switch on. Hopefully you can hear that the low end of the snare drum is considerably enhanced with the polarity switch on, and we've got back our fundamental vibration frequency that was lost due to cancellation. Now over the years the polarity switch often became renamed as a phase switch. Which comes across as a slightly more sophisticated term for engineers to use. But phase is the wrong word to use really, since phase cannot be switched or flipped. It can only really be shifted in scientific and mathematical terms. But the term phase is now part of our language in audio recording. Rightly or wrongly, it's here to stay. But I still believe it's more useful to think about time delay and signal polarity when trying to solve a frequency cancellation problem in the studio. So it's always important to check for phase coherency or frequency cancellation especially when recording drums. Often this means taking time to listen to the individual closed drum mics when mixed with the overheads. You may find that cutting out higher frequencies can help you hear in the control room to evaluate if the lower, more powerful frequencies are being cancelled. In addition, we sometimes reduce some of the lower frequencies in the overheads at the mixing stage. So frequency cancellation can be less of an issue with the kick drum than with the snare, which has a higher fundamental frequency and is significantly represented in the overheads. It's also worth checking the overheads, even though they'll be panned left and right in the mix usually. When we sum them to mono in the center, sometimes it's possible to hear that important frequencies get cancelled. And you'll need to check drums which have more than one close mic too. Often as we have more than one microphone recording, say the kick drum, and these can easily cause frequency cancellation when mixed together too. But there's no perfect solution to frequency cancellation. It's impossible to avoid and becomes even harder to achieve with the more microphones you add to the setup. But it's certainly worth spending some time evaluating at the start of a recording session. Because achieving perfect alignment of the most powerful drum frequencies can make a big difference in the clarity and impact of your recordings. Now that we've discussed the topics of phase coherency and comb filtering, we can move on to discuss more practical aspects of drum recording and production. One of the first things we need to do when planning a drum recording session is to choose a suitable room to record in. And it's quite amazing how often the issues related to phase and comb filtering influence our choices with regards to a drum recording session. And in particular, the way in which the instrument interacts with the recording space is of great relevance to the accuracy of the recordings that we capture. Although there is some subjectivity in this, in general, I would argue that larger rooms work best for drums, for a number of reasons. Firstly, larger rooms mean that any reflections to the microphones from the walls or the ceiling will have travelled quite a long distance. This means they will be at a much reduced volume than the direct sound from the drums themselves, and any comb filtering that occurs will be quite minimal. In comparison, smaller rooms can cause significant frequency cancellation, owing to the great amount of comb filtering caused by reflections from close walls or a low ceiling. And this is why drum recordings made in small rooms often have a sound that people describe as boxy. Equally, the longer distance to the walls or ceiling in a large room also means that a longer time passes before the sound reflections arrive back at the microphone. This has a valuable effect on a recording by providing a very clear differentiation in the sound that is directly from the instrument, and that which is reflected from the walls. Furthermore, you probably know that all rooms have their own frequency modes, which result in some sound frequencies being enhanced, especially between parallel walls. Well, the room modes of smaller rooms tend to be at frequencies inside the human hearing range, and at frequencies which coincide with those of many conventional musical instruments. However, larger rooms have much lower frequency modes, sometimes below our threshold of hearing. Equally, there also tends to be a weaker mode coupling between parallel walls that are 10 or 20 metres apart, meaning that the energy of a room mode dissipates before it has chance to really build up and become troublesome. Now, quite frequently there is one space in the performance room where the drums just sound a bit better than anywhere else. Usually it's in the middle and fairly central. But often a slight off center placement or an angle to the walls can have a positive effect on avoiding any room mode peaks or nulls that you might experience. A good practical suggestion to find the best position to place the drums is to walk around with a floor tom and a soft mallet. Hit the drum as you walk around and listen to the response in different locations. You might hear that some locations sound too boomy, and other locations sound to have less bass. But hopefully you'll be able to find the location where the drum sounds both powerful and clear at the same time. Once you have decided where to position the drums, you'll obviously need to decide how many microphones you want to use for recording the drums, and where to place the microphones. And there are literally hundreds of different microphone techniques. And my advice is to research and try out as many as you possibly can. It's really quite possible to get great drum recordings with just three or four microphones. And I also recommend anyone to experiment with recording drums using just a few microphones, as a learning exercise if nothing else. However, when I'm recording drums, I tend to place close mics on all the drums, and the cymbals too, which allows me a lot more creativity at the mixing stage. I find that when mixing, I sometimes want a very clear ride cymbal in a section of a song. I also often find that the hi hat sound becomes compromised by any treatment applied to the snare drum mic. So a close hi hat mic allows me more control of that sound and its positioning when mixing. A main point of attention when drum recording is considering the strategy for stereo placement, and that helps determine what type of overhead microphone technique you might use. I find there are a few stereo concepts to consider when producing drums. Firstly, for each project, I want to decide if the drums shall be reproduced fairly authentically, As if the recording captures the sound of the instrument, quite naturally, and as it sounded in the space that it was recorded. Or, do I want the drums to sound more hyper real on playback? As if the listener is literally sat in the middle of the most incredible drum kit, which has power, a snap in the snare, exaggerated wide cymbals, and drum fills that roll from left to right and vice versa too. Both the authentic and hyper real approaches are good and viable, and each approach suits different music genres and projects too. So there are two things to consider here, which both relate to the listener perspective of the stereo field. Now if I sit down and play a drum kit, I hear the hi-hat clearly over to one side, the ride symbol over to the other side, the kick in the snare, fairly central and the tom spaced around between the two wide symbols. The same applies for piano. If I sit down and play, I see the base notes performed by my left hand and the treble notes on my right hand side, and the sound predominantly follows this stereo spread too. But if I sit in the audience and listen to someone else play the drums or piano, I get no concept of this stereo positioning at all. All I hear are the instrument as a fairly mono source sound coming from the stage, and some stereo information related to the room and the reverberation environment that I'm sat in. So clearly both of these scenarios are common and acceptable in music production, and the music producer gets to decide if they want the listener to experience the hyper stereo version of the performer, or the sound and space version, as we might call it, that the audience experiences. Equally, we could find a midpoint or a hybrid version of these two approaches. And this consideration of the stereo perspective can help inference what type of overhead microphones you might use in a drum recording session. Now you may know already that there are many stereo techniques for recording instruments, and the drum kit is an instrument that has sound elements positioned across the stereo field from left to right, perhaps more than any other instrument. So stereo techniques are really important for drums. And I'll give a quick overview of some common, and not so common, overhead recording techniques and methods now. And we'll listen to some examples of a few different approaches as we go through them too. One of the most common approaches to overheads is to use the spaced pair technique. This literally just means spacing two microphones above the drum kit, with one positioned to the left, and one positioned to the right. Usually the microphones are one or two meters apart. Conventionally this technique uses omnidirectional microphones, but directional microphones can be used too. The space pair technique relies on the time delay in sound reaching one mic before the other, which is relative to the instrument positions in the performance space. So obviously a cymbal on the left hand side reaches the left microphone a few milliseconds before it reaches the right microphone, and our brains are capable of interpreting this as stereo positioning when we listen back. In stereo, there isn't usually a huge issue with comb filtering, because we position the recordings to the left and right channels in mixdown. But be careful to check mono compatibility here, because when the left and right channels are summed together, there can be comb filtering or phase cancellation occurring. A good modification of the space pair technique is the Glynn Johns technique, which is named after the man who developed the method. The Glynn Johns technique has one of the overhead microphones spaced high above the snare and hi hat, and the other microphone positioned lower down in the region of the floor tom. It's important to measure the distance of each mic from the snare, and set up both equally. This ensures that phase coherency and an authentic snare sound can be achieved. Usually a kick drum mic is added too with the Glyn Johns technique. This gives definition and covers any phase issues that could occur with the kick drum from the spaced pair. And it's acceptable to add a close mic to the snare too if you want more control over that sound at mixdown. Let's take a listen to a drum recording with the Glyn Johns technique. Using Neumann U87s as overhead mics, an AKG D12 VR on the kick drum and an SM57 on the snare. Let's hear that with just the overheads first. And now we'll add in the close snare and kick mics. In the mixdown you can hear a strong overall sound of the drums and a wide stereo positioning of the toms using just 3 or 4 microphones. Another common overhead technique is the coincident pair method. This means we have two directional microphones positioned at exactly the same location, but pointing in different directions. The simplest version is the XY technique, usually using small diaphragm cardioid mics, angled at 90 to 120 degrees apart. So one microphone points towards the right side of the drum kit, And one points to the left hand side. Whereas the Space Pair technique uses time difference to differentiate between left and right signals, the Coincident Pair technique uses the fact that the directional microphones capture more sound energy, or volume, from the source they are pointing at, and less sound from off axis positions. This is great because there's no issue with time delays or comb filtering when the Coincident Pair is summed to mono. On the downside, the center of the kit, where the kick and snare are positioned, are somewhat off axis to both microphones, and the center image can be less defined with an XY pair recording. For this reason, it's almost essential to add close mics to the kick and snare, and possibly the toms also, to get a good overall sound of the kit. Two modifications of the XY technique are the Blum Line Pair, which is identical but uses ribbon mics and a figure of eight pattern, and the ORTF method, which is a personal favourite of my own. The ORTF technique is named after a French radio station that pioneered the method. It's essentially the same as the XY approach, but with the microphones also spaced a short distance apart, around 30cm. ORTF is more correctly termed a Near Coincident technique, because it brings some of the benefits of both the Coincident and Space Pair techniques together. So let's hear that same drum kit which we listened to before, but this time with Neumann KM184 mics as ORTF overheads, and using the same kick and snare mics as before. As before, let's just hear the ORTF overheads alone first. And now we add the close snare and kick mics. The ORTF method gives an excellent reproduction of the cymbals and sparkle to the kit, but it does need the close mics on the kick and snare to give a good overall sound. Close mics on the toms would be useful too in order to get definition and a clear sense of stereo positioning in the mix. Another important stereo microphone technique is the mid side setup. This method aligns more with the earlier discussion around balancing a blend of direct and spatial sound rather than aiming for a hard positioning of sounds from left to right in the stereo field. With mid side we use two microphones. One is a cardioid mic pointing towards the middle of the drum kit. And the other is a figure of eight microphone, pointing perpendicular to the sides of the kit. The principle is that the center is captured by the forward facing mic, and the two sides are captured as positive and negative sound pressure on the front and back sides of the figure of eight mic. But with mid side, we need to use a little post production after the recording to achieve a true stereo signal. So let's just talk through the post processing required to turn two mid side microphone recordings into a true stereo signal. In post production, we need to copy the figure of eight, or side recording, and invert it with a polarity switch. In most DAWs, you can invert a signal with either a plug in, or with some offline processing. We then pan the original side recording to the right, and the inverted version to the left, or vice versa, depending on how you oriented the microphones. When the mid and two side signals are mixed together, we then get a stereo reproduction of the instrument we recorded. What's fascinating with mid side is that you can have just the mid mic, and then choose how much side you want for taste. Let's hear that. In this recording I used AKG C414 mics as mid and side mics. Let's first hear the mid microphone alone, which is essentially a mono recording of the drum kit. Now we can fade up the side channels to hear the playback become stereo. Now let's also add the closed snare and kick mics to the mid side setup. You can hear the sense of space develop as we introduce the side channels, and the recording really does come to life. There isn't as much perfect left to right stereo in the playback, but we get a superb balance of direct definition and of the space which the drum kit was recorded in. Let's just pause here and take another listen to the three different overhead techniques and compare them together. First we had the Glyn Johns recording, then the ORTF recording, finally the mid side recording. Do you have a preference? Well, to me it really depends on the project and the genre. If I were recording folk or acoustic music I would probably go for the mid side method. Punk or lively rock music, I would probably use the Glyn Johns recording. And for something more commercial such as rock, pop or metal, I'd use the ORTF with the additional close mics on the toms. But there are no rules here, it all depends on personal preference and a sense of how you want to mix the track to get the desired final result. It's quickly worth mentioning a few other interesting techniques that you might want to research and try for yourself. These being, the Decca Tree method, which uses three spaced pair mics and was used for many years as the go to classical recording method. The baffled omni pair technique, which mimics the shadowing of a human head between two microphones positioned as ears. In fact, you can even buy head shaped stereo microphones, which are surprisingly effective for drum recording. And finally, the left right center technique, which is essentially the XY or ORTF approach, but with a third microphone pointing towards the center, allowing an improved center image in playback. It's really important to experiment with all of these techniques, and make your own judgments about which suit you and your methods best. There really is no other way than to experiment and hear the results for yourself. Equally, you should experiment with different microphone types and models, especially when recording snare and kick drums. There are many articles out there on what microphones and techniques to try out. In all cases, it's also worth experimenting with slight adjustments to the positioning or the angle of the microphone. Even small changes in positioning can have a big effect on the sounds that you capture. For example, it should be no surprise that if you point a snare mic towards the center of the drum, you'll pick up more of the drum's fundamental frequency. Angling the mic towards the edge of the drum will pick up more of the drum's overtones, which you may or may not prefer. So that completes this podcast on drum recording techniques. We've taken a deep look at how the science of sound affects the recording of a complex instrument such as the drums. And we took that knowledge and used it creatively to explore different techniques for capturing exactly the types of sounds we want for a particular project. I hope you have fun experimenting with drum production and developing your own personal recording methods and techniques too. I'm Professor Rob Toulson and this has been an RT60 limited production for Sound On Sound. Thank you for listening and be sure to check out the show notes page for this episode where you'll find further information along with web links and details of all the other episodes. And just before you go let me point you to the soundonsound.com/podcast's website page where you can explore what's playing on our other channels.