Binaural recordings are an easy and effective way to capture immersive audio.
Binaural audio is one of those techniques that the recording industry seems to rediscover every decade or two. It’s actually as old as the hills (some of the earliest experiments date back to the 19th Century!), but because binaural recordings need to be heard on headphones to get the full effect, it has had limited applications. However, as headphones have become the dominant consumer listening medium, it’s more relevant now than ever.
Binaural recordings are stereo recordings that take advantage of the mechanics of human hearing. Our eardrums work an awful lot like pressure‑operated (ie. omnidirectional) microphones, and since we’ve got one on either side of our heads, our hearing is essentially the same as a spaced‑omni stereo mic array. Now, unlike coincident mic arrays such as X‑Y or Mid‑Sides, in which sounds always arrive at the two mics at the same time, spaced arrays rely on the possibility of sound arriving at the two mics at different times in order to convey positional information. So, if a sound being picked up by a spaced array (or our ears) arrives at both mics (or eardrums) simultaneously, then we know it’s coming from directly in front of or behind us. If it arrives at the left mic or ear slightly earlier than the right, then we know the sound is coming from the left, and so on.
But the human hearing system is rather more advanced than a simple spaced stereo mic array, not least because it can resolve sounds in three dimensions, rather than just the 2D plane of stereo audio. One of the ways it does this is using those little radar dishes stuck to either side of our heads. Our outer ears, or pinnae, ‘funnel’ sound into our ear canals, filtering it in complex ways as they do so, and that filtering changes depending on the angle at which sound arrives. This is one of the ways we can resolve the ambiguity about whether a sound is arriving from the front or rear, or above or below, even if that sound arrives at both of our ears at the same time.
Our heads themselves also have a big effect on how we hear. Our skulls are relatively large resonant cavities, with more resonant cavities (sinuses, the oral cavity and so on) inside them. Being quite large, they also form ‘acoustic shadows’, occluding higher frequencies if the head prevents sounds from reaching either ear directly. (This acoustic shadowing phenomenon, incidentally, also forms the basis of the Jecklin disc, which is similar to a binaural recording setup except that it’s designed to be heard on speakers as well as headphones.)
The combined acoustic effect of a person’s head on their hearing is called their Head Related Transfer Function, or HRTF, and everyone’s HRTF is different. People’s outer ears are as unique as their fingerprints, and of course our heads all differ in shape and size. This means that the effectiveness of binaural recordings — how precisely we can locate sounds when listening to them — can vary from person to person. There are two ways around this, the most established being the use of a ‘dummy head’.
The inscrutable chap you see at the top of the article is called the Neumann KU 100, and everything about him — his size, shape, ears and chiselled features — has been mathematically modelled to represent the average adult human head. Inside each of his ears lies an omnidirectional microphone, and the idea is that, by using a head with attributes that lie in the middle of the range of humans’ (ear height, inter‑ear distance, head density and so on), a recording made using the KU 100 should sound right on the widest range of listeners.
Danish measurement experts Bruel & Kjaer take the idea even further with their HATS (Head And Torso Simulation) range. These dummies comprise a similarly contrived dummy head plus an artificial torso, for even greater realism (the chest and shoulders being two other sources of resonances, absorptions and reflections, of course).
Those two examples represent the pinnacle of binaural dummy head design, and they’re also fantastically expensive: either will set you back many thousands of pounds. Happily, though, there are some cheaper options available, among them the fairly new 3Dio FS series. These are dedicated binaural rigs that do away with the ‘head’ and comprise just a pair of silicon ears, on a bar that holds them apart at a suitable distance. They’re available at a range of prices, from the premium FS Pro II, which uses DPA omni mics and has XLR outputs, to the affordable FS, which uses cheaper electret capsules and connects via mini‑jack to your phone or tablet. I’ve been playing with the FS Pro II during lockdown, and my experiments with it were actually what inspired me to write this article. I’ll be reviewing it properly in an upcoming issue of SOS.
But you don’t necessarily have to buy a dummy head, or even a creepy‑looking pair of prosthetic ears, to start experimenting with binaural recording. After all, why reinvent the wheel?
It’s perfectly possible, and sometimes even useful, to be your own dummy head. If you’ve got a pair of miniature omnidirectional microphones and can contrive some way of fitting them in front of your ear canals, then you’ve already got a binaural setup and you’ve saved yourself the trouble of having to dig out a microphone stand. If not, there are a number of headset kits available that include the necessary mics (almost always electret mics, since those are the easiest to make small enough) and mounting hardware (ear‑hooks, usually, much like those used for hearing aids or in‑ear monitors) to get you started. You can pick up a set very cheaply: Hooke Audio’s Verse system is affordable, and also records via Bluetooth straight to your phone, as does the Sennheiser Ambeo Smart headset, though the latter only works with iOS devices. For a more professional solution, you could opt for something like DPA’s Binaural Headset, which uses their high‑quality 4000‑series capsules, but which would obviously require an audio interface or portable recorder.
Being your own dummy head has advantages and disadvantages. In the ‘pros’ column, a binaural recording made using your own head will always sound right to you when you listen back to it on headphones, because it was subject to the same HRTF that you use for listening to sounds every day. It’s also a relatively discreet setup: should you want to discreetly capture a live performance from the audience’s perspective without punters wondering what Kryten is doing at their open‑mic night, self‑worn mics are the obvious way forward.
Down sides include the fact that any noise you make during the recording will definitely be picked up (so no sneezing, yawning or tooth‑grinding during the performance!), and that your legs might get tired with all that standing around. You’ll also need to be very careful about moving your head during takes, because that motion will be captured in the recording. Perhaps most significantly, though, there’s a chance that your HRTF is sufficiently different from the listener’s that the glorious sense of three‑dimensional space you can hear on the recording simply won’t translate. I seem to have a fairly average head, because I tend to find binaural recordings made on all kinds of setups immersive and effective. One of my colleagues is not so lucky, though: for him, vertical information and front‑back ambiguity are a big issue, and it follows that if he were to make a binaural recording using his own head, there’s a high chance it wouldn’t translate very well for other listeners.
When deciding where to put a binaural array, there is no better place for it than wherever your ears are when you’ve found the best balance of instruments and acoustics.
So now you know how binaural recordings work, and how to make them. But why would you want to? Well, if you subscribe to the notion that a recording should aim to capture as closely as possible the sound of an event itself, there’s a purist argument for only recording concerts and recitals binaurally. But there’s also a very practical argument to be made, concerning mic placement.
In last month’s ‘How To Mic Anything’ article, Paul White cautioned against simply putting up microphones where the sound in the room is best, because “microphones don’t ‘hear’ the world in the way we do”. That’s very true if you compare a cardioid microphone to the human auditory system, but a pair of omnis in a dummy or human head can get pretty damned close. When deciding where to put a binaural array, there is no better place for it than wherever your ears are when you’ve found the best balance of instruments and acoustics.
In terms of practical applications, audio for video is high up the list. One of my most successful experiments with the 3Dio kit was to rig it up on top of a DSLR camera, and use it to capture sound while filming. The precise nature of the sound localisation, combined with the visual reinforcement from the footage, made for a satisfyingly congruent whole, so if you’re trying to capture a natural sound while shooting a video, I can highly recommend a binaural setup.
Foley is another obvious application, as is the recording of sound effects for video games. A huge number of games use a first‑person view, and if the player is visually sited inside the character’s head, it makes sense for sound effects to be recorded from the same perspective.
Finally, while researching this article, I found out from our Technical Editor Hugh Robjohns that a great many of the BBC’s radio sound effects were recorded binaurally — although not necessarily because they were intended to be heard on headphones. Many people listen to the radio in mono, and the relatively close nature of the mics in a binaural setup means that recordings made with them are inherently quite mono‑compatible. And when heard on a standard stereo speaker setup, although the effect isn’t quite the same as on headphones, binaural recordings still provide a useful bit of stereo width. It is, after all, just another stereo mic array!
If you’re a Logic user, you may have noticed a new binaural panning option appear in the mixer a couple of years ago. Binaural processing, in essence, ‘converts’ a normal mono or stereo recording into a pseudo‑binaural one, by applying inter‑channel delays and filters to replicate the effects of a human head.
In the case of Logic (and also Sennheiser’s free Ambeo Orbit plug‑in, which works in much the same way), this means you can pan sounds around, in front of, behind, or even above or below the listening position. Like a true binaural recording, this requires headphones to work properly, and will be subject to the vagaries of your personal HRTF, as mentioned in the main article.
Binaural audio is also used in ‘virtual control room’ plug‑ins such as Waves’ Abbey Road Studio 3, Embody’s Immerse Virtual Studio, dSONIQ’s Realphones and Slate’s VSX. These use impulse responses, either captured with a binaural rig or subsequently processed binaurally, of a high‑end monitor setup in a professional control room. Pipe your DAW output through them and listen on a pair of headphones, and hey presto! You’re mixing your band’s demo in Abbey Road, or Ocean Way, or even (as in Slate’s VSX) through a car sound system.
Finally, if you’ve got an Ambisonic microphone such as Rode’s NT‑SF1 or the miniDSP AmbiMIK‑1 reviewed elsewhere in this issue, you can convert recordings made with it into binaural ones. As with a binaural panner, this is done by applying filters and delays to the left and right channels, and while these will only ever approximate the effect of a proper spaced‑mic binaural array (because Ambisonic mic capsules are coincident), you have the advantage that you can ‘steer’ the recording after the fact, essentially changing the direction in which the simulated head is facing.