Setting off in the cause of research…or seeking ridicule?
I think Lockdown III must be getting to me: I’ve been pondering over stereo ambiences for film lately, wondering if they ever are or should be recorded while moving to match the POV (point of view) of the camera?
This was stimulated by thinking about sound for an upcoming project with drone footage, where the drone will follow the course of a small local river from source to sea – flying fairly low and slowly. I want the sound to match the view – along the bubbling stream, the passing over of weirs, cows and sheep being flown past etc. Obviously bunging a couple of mics on the drone isn’t an option given the noise of the motors and rotors, so any ambiences will need to be recorded quite separately. Initially I thought of helium balloons: I calculated that seven normal-sized party balloons would lift a pair of Rode Wireless GO II transmitters into the air, but, the more I have thought about this the more complex it seems: a totally windless day seems essential and, even then, trying to stop the mics spinning or, harder, trying to get the balloons in the right place seem fraught with difficulties. And then, more fundamentally, what does a moving stereo recording actually sound like? In other words, would it even be worthwhile trying to do this?
Recording to the miniscule MixPre-3
Drawing a blank through internet searches (the nearest thing being people trying to record the sounds of – say – a bike travelling, rather than just the ambience without evidence of the mode of transport), and not knowing how the soundtrack of any film with apparent moving stereo ambiences was actually made, it seemed easiest to experiment. A bicycle appeared to be the best bet for near-silent travel, as long as fast speed or free-wheeling was avoided. After initial less than satisfactory attempts with semi-binaural set-ups of lav mics by my ears and either side of a rucksack (both getting far too much bike noise, and, also, showing up the inadequacies of furry wind-protection for these mics), I rigged up a boom pool to a sheet of ply, cut to the shape of a rucksack, to mount the mics well above me (so further from the bike and road noise) and to prevent unwanted rotation. The mics were omni SDCs mounted end-to-end in a single blimp to minimize windnoise (see my post here on using mics in this way). Extending the boom pole very far got unwieldly, so it was only partly extended. Sound was recorded to a Sound Devices MixPre-3 on a light harness. The results were in a different league to the lav mic experiments, but the bike was still very evident and, above all, I’m not at all convinced that the sense of movement is very strong. This last point is critical. In short, I have no doubt that a better result could be achieved by recording general ambiences and specific sound effects from static positions and amalgamating them into the soundtrack to give the illusion of moving through space: doubtless this is what sound designers know and do, anyway, but it is good to experiment and find that, sometimes, the seemingly logical approach doesn’t work. And if, in doing so, I’ve gained an eccentric reputation in the village (yes, cycling with a blimp on a vertical boom pole does look extremely silly), then so be it…
For anyone interested, here is a recording of one of my tests with the SDC spaced pair in blimp, including the reaction of a passing neighbour.
Rode Wireless GO II recording into an Android phone
Putting your equipment through its paces is part of getting to understand it properly. Given its newness (only released this week), my tests on the Rode Wireless GO II might be of wider interest: so here’s a post about them.
First, a quick bit of background. I’m not normally in a rush to adopt new technology, but this week – following the death of an old handheld recorder (my Sony M10) and while planning a couple of projects – I was looking at lav mic self-recording back-up options, and Rode’s new offering seemed on the nail and very timely. I know the first version was nothing like a professional wireless mic set up (not least for its dropouts when out of line-of-sight), but I’d found it eminently usable for the particular and less exacting uses I bought it for: in particular, the sound quality stood up compared to the same lav mic hard-wired. So, rather than wait for anyone to get round to a thorough test – as opposed to the numerous unboxing and product review vlogs that will doubtless flood YouTube – I ordered one from CVP.
Overall audio quality
There are different aspects to the audio quality of such a system, including its on-board mics vs external lavs, its on-board recorder vs transmitted audio, and both the recorder and the transmitted audio vs a better system.
Using a Sound Devices MixPre-3 to provide a decent sound source into the Rode Wireless Go II transmitters
Most interesting to me was to take the on-board and external lav mics (both of which have relatively high self-noise due to their small diaphragms) out of the equation, and test the recorder and wifi alone. For this, I fed the two transmitters with a stereo signal of a simple acoustic guitar recording (recorded with a spaced pair of 7dBA Rode NT2a mics in fig 8 mode) via my Sound Devices MixPre-3 (feeding a signal from the stereo out at -6dB to give a reasonable level at the transmitters, comparable to that of the mics above). I recorded the transmitted signal via the USB out of the receiver. Here are the clips, with no processing added:
Original MixPre-3 recording:
Rode Wireless GO II on-board recorders in uncompressed/broadcast mode:
Rode Wireless GO II on-board recorders using the compressed/standard mode:
Rode Wireless GO II transmitted recording:
The verdict? Well, see what your ears say. The most obvious positive is that the Rode Wireless GO II on-board and transmitted audio show remarkably low self-noise: there is effectively no hiss in the short lead-in. The most obvious negative is the lack of bass response: OK the datasheet shows a frequency response graph with a sharp fall-off below 100Hz, but this appears to relate to the mic. Needless to say, you can boost the bass response in post, and here is a quick EQ’d version (I looked at the bass fall-off in the Rode Wireless GO II using pink noise, though my resultant EQ might be a bit too heavy – but it makes the point):
And, of course, what is effectively a high-pass filter isn’t a significant problem with intended use (speech) and will aid reduction of wind noise and other unwanted low frequencies.
So my overall view on the audio quality of the electronics? Well, the transmitted and on-board recordings sound surprisingly good: OK not up to the level of a Sound Devices MixPre-3 and not ideal for music recording, but easily good enough for use with the on-board lav mics or external lav mics (or other PIP mics such as Micbooster’s Clippy mics) for speech or ambience/nature recording, and perhaps usable – in the way that you might use a small handheld recorder – for music recording where, for whatever reason, a better and more conventional set-up wouldn’t work. Also, the compressed on-board recordings don’t sound too bad at all, despite my natural dislike of lossy compressed formats!
On-board mics
Given the mic in the original Rode Wireless GO, I was reasonably optimistic about it and, also, conscious that a) different people like different lav mics and b) I don’t own any of the more expensive lav mics (such as those made by DPA). But, for what it is worth, here’s a simple comparison of Rode’s lav mic (that’s their more expensive one – at c.£160 – and perfectly usable) vs the on-board mic recording a voice (apologies for the performance!), and recorded via the Rode Wireless GO II’s receiver (both digital out, into an Android phone running USB Audio Recorder Pro, and via the 3.5mm output into a Sound Devices MixPre-3).
Rode Wireless GO II on-board mic via digital out on the receiver:
Rode Wireless GO II on-board mic via 3.5mm analogue out:
Rode lavalier mic plugged into Rode Wireless GO II via digital out on receiver:
Rode lavalier mic plugged into Rode Wireless GO II via 3.5mm analogue out:
The verdict? The on-board lav mic is quite usable if – and this is the key – you don’t mind the transmitter being visible too. The much more secure furry windshields of the new model at least make this more feasible. And as for the difference between digital and 3.5mm output, there is little in it. Well at least with a decent sound recorder: of course, the difference will be greater if choosing between USB into a phone and analogue into a camera with poor audio.
Range and dropouts
As I said, I found the original Rode Wireless GO usable for my purposes, but I can’t deny that a more robust wifi signal wouldn’t be useful. Indeed, for many, it would be essential. The 200m line-of-sight spec for the new version (vs 70m for the original version) shows an improved wifi capability, and this is borne out even in initial, simple, testing. For example, whereas the original Rode Wireless GO – under my test conditions (a rural Norfolk village) – experiences dropouts from around 20m when worn on the rear of the subject (or ‘talent’) – i.e. blocked by the wearer’s body – the new version worn in the same way is good for about 35-40m. Obviously, different locations will produce different results, but this suggests much more suitability for, say, wedding videographers, where, of course, the recorder provides a back-up anyway. For professional sound recordists’ use (e.g. television and film) the improved signal strength is probably immaterial as the units lack the range, features, robustness and ease of control of professional wireless systems such as those produced by Audio Ltd. and Lectrosonics.
Field-recording (an update of 15.3.2020)
Rode Wireless GO II TX units in action for stereo ambience recording…
Having had the Wireless GO II for a few weeks, I’ve had a chance to use it for different scenarios, one of which is as a remote field recorder. Here the ability to use the two TX units as a remote spaced-pair of omni mics opens up all sorts of possibilities, not least recording sounds of nature (such as timid birds) from afar. Of course, if the tests above suggest that the on-board mics or the electronics aren’t are not high enough quality, you can simply use the system for monitoring: for example, running SDC mics into a Sound Devices recorder, and plugging the TX units into the recorder’s stereo out so you can hear what is being recorded from afar. Taking the Sound Devices example you could, of course, use the wireless remote control (Wingman) for the recorder too, though I haven’t tried that combination or, indeed, tested the Bluetooth range. Anyway, what I have tried out is simply recording remotely with the on-board mics, both within a Rode blimp and with their supplied fluffy windshields: of course, the former, while restricted to 35cm spacing, provides more wind protection, but the Wireless GO windshields on their own were surprisingly good and allow any spacing (in my tests this happened to be 46cm). And the fluffy windshields of the new TX units fit much more securely than those of the original Rode Wireless GO. So here are a couple of examples, one in very high wind (we had gales at the weekend) and one in slightly less windy conditions. Both just with the TX units alone and their own fluffy windshields, and simply placed in a holly tree in the garden (near the road), as in the photo above. The recordings are from the RX unit’s USB output.
Stereo ambience recording (garden in high wind):
Stereo ambience recording (garden in moderate wind):
A pair of the tiny Primo EM272 omni capsules: add two wires and a 3.5mm plug and you are ready to record!
Primo microphone capsules are much-loved by the DIY crowd, especially the omni EM172 (now replaced by the EM272) capsule: it is a 10mm capsule with 14dBA self-noise. As such it offers similar scope to a lavalier (lav) mic for times when you want something much smaller than a small diaphragm condenser (SDC) mic, but has much lower self-noise: lav mics typically have around 22-5dBA self-noise, which can be pretty noticeable for many recording uses, such as ambience and nature recording. The sound quality is on a par with capsules used in mid-range handheld recorders, but, of course, once you have external mics you can be much more flexible. For example the omni mics on the much-loved Sony PCM10 are too close together for stereo, so plugging in a pair of EM272 mics allows for a proper spaced pair: you can clip those to your hat, the sides of your specs or, to look slightly less silly and to avoid your head movement giving odd shifts in the stereo image, to the sides of a bag or rucksack.
And the best thing about the Primo capsules is that they are cheap. Currently in the UK an individual bare capsule is £12.78 from FEL Communications Ltd (Micboosters): FEL/Micboosters also offers various versions (e.g. pre-soldered), other Primo capsules (including cardioid and figure 8: see below), matched pairs and, for those not into DIY, capsules already made up into finished mics – their Clippy and Pluggy models. There are a couple of other manufacturers who use Primo capsules in affordable mics, the most well-known of which is LOM in Slovakia, but I think Micboosters is the only one that also sells the bare capsules: and it is one of those great small British companies, run by the very helpful Nick Roast, who has worked as a BBC sound engineer for over 30 years.
A pair of Clippy mics
I have a pair of the Clippy mics (together with Rycote furry covers that are made specifically for them), which are useful for discreet or minimalistic stereo recordings and as lower noise lav mics for dialogue. And I have used bare EM172 capsules for what Curt Olson – who inspired my experiments – calls a ‘head-spaced parallel barrier array’, albeit in my case small enough to fit inside a Rode Mk1 blimp and with some of the mic placement attributes of SASS arrays: the circular baffles are c.90mm diameter and the mics are 160mm apart. Surprisingly effective. I’ve also used an EM172 to make a boundary mic, with the capsule set off-centre in a disk of perspex (150mm diameter and 5mm thick). Perhaps next I should buy some silicon ears (I see Micboosters sell them too) and make a binaural head with a pair of EM272s, as others have done.
Primo also make a single diaphragm figure 8 capsule (the EM283), again 10mm diameter, but this is not normally available via retailers. I noticed that Micboosters had started selling them and, as I had never heard of the capsule nor could find anything about it via the internet, I bought one for the princely sum of £19.68 for fun/curiosity. The specs are not as attractive as the EM172/EM272 and the 22dBA self-noise might be rather too high for many, but it is fine when used as part of a mid-side pair for louder sources such as music or some street ambiences. I found that the EM283 capsule needs better RFI screening than the EM172 and EM272, but that’s nothing unusual. I just bunged it in an old sawn-off shotgun mic tube for testing and, though this makes the mic unnecessarily large, it is fine – all hum eliminated. At some point I’ll get around to making a smaller housing with the fine mesh screening I have bought for the job. Not entirely sure if I’ll make much use of this capsule, unlike the EM172 and EM272 ones, but it has proved useful in an odd way: I had been thinking of acquiring an AKG CK94 figure 8 mic to provide the side mic for a mid-side pair with my CK93 hypercardioid, but was worried about self-noise (the CK94 is also 22dBA) and this convinced me that for my intended use I really do need something a lot quieter (so will need to save for an MKH30). UPDATE 26.5.2021: despite my conclusions about the EM283, I did buy a used AKG CK94 after all (a bargain came along) and its theoretically identical self-noise of 22dBA is not at all problematic. Lesson learned? That not all self-noise specs are equal! I’ve written a blog post about the new mic here.
So the final word: I’d really recommend playing around with bare Primo capsules as a cheap way to learn about arrays, and, even if DIY isn’t your thing, I’d recommend a pair of ready-made Clippy mics with these capsules as a great and very affordable alternative to a pair of lav mics for those occasions when SDC mics (and P48 power) aren’t feasible.
Primo EM272 capsules as delivered: matched pair with measured sensitivity a bit better than the published specs.A head-spaced parallel barrier array – with the influence of SASS arrays – built to fit inside a Rode Blimp.A boundary mic with an EM172 capsule: so simple and cheap, but better performing than so many commercial offerings.A size comparison of a Clippy mic and a miniature lav (in this case my Rode lav). The Clippy’s EM172 capsule makes it rather chunkier, but for many uses this isn’t a problem, and it is still small enough to hide under much clothing.Experimenting with a Primo EM283 fig 8 capsule: mounting it inside part of an old shotgun mic body.And using the Primo EM283 fig 8 mic as the side mic in a mid-side mic pair with a Rode NT55.
Though not strictly part of filmmaking or sound recording, ultimately playback is on hi-fi speakers (well for those not limiting themselves to earbuds!) and having something decent on which to check mixes other than nearfield studio monitors is critical. To me, speakers means DIY and I have been playing around with them since childhood: an early speaker project around undergrad days was a long-throw horn using WEM drivers, and then a few years later, when a thick cold meant I wasn’t feeling much like working on the thesis, I made my first exponential tapered horn speakers, loosely inspired by Laurence Dickie’s famous Nautilus B&W speakers. The rationale of the Nautilus design (and subsequent speakers using this approach) is to use tapered tubes to absorb the rearward sound produced by the drivers, to stop it being reflected back (as in a conventional speaker cabinet) through the driver again, adding unwanted resonances and colour.
The 1990s old snake speakers, with B&W drivers and that chipboard aesthetic
These first tapered horn speakers weren’t snail shaped like the Nautilus (too complex and derivative) but, rather, vaguely snake shaped, built simply out of chipboard, and had more of the WEM PA speakers: they weren’t very hi-fi at all. But modifications quickly followed. I was given some spare B&W drivers (I think for their 802s, but they are marked ‘prototype’ on the back), so modified the snakes to take them, and then some backroom boffins at B&W kindly stuck them in their anechoic chamber, stuffed them properly (I was too tentative with my wadding), added plasticine under the tweeters to take out a nasty reflection, and designed and built crossovers to suit: this might seem unusual, but I was brought up in the small Sussex market town of Steyning, where B&W’s R&D centre was located, so enthusiastic and generous research staff were thick on the ground.
I’d always meant to house the speakers in something rather nicer than chipboard, but the decades slipped by (actually over 25 years) and it took the Covid-19 lockdown for me to get round to it. Somehow the project rather developed: instead of re-housing the early 1990s B&W drivers (or, actually, similar: I had a spare set), I was offered some rather more modern and better (indeed, world-class) drivers (including metal-coned pistonic bass/mid drivers) actually designed for this type of speaker, and, again, some technical support (e.g. on crossover design and build). I’ll draw a discreet veil over which (different) manufacturer gifted them this time (those interested can probably guess), as they don’t supply for DIY use: again, these were some spares knocking around – acoustically fine, but not quite pristine looking.
Half-model for the new design
For the new speakers I decided to go for laminated plywood construction. I like the aesthetic of varnished laminated plywood sculptures and it seemed an achievable way of building a complex curvaceous shape: there would be no flat sides this time round. Also, with laminated construction I could create fairly thick and inert walls. The overall form is similar, nonetheless: a rearing snake (or a large number 2 if you want to be more prosaic) and a two-way design with passive crossovers. And the underlying principle was the same. That is, the form was defined entirely by eye and a gut feeling that the exact taper of the tube isn’t critical, so there is no scientific analysis behind it! Work began with drawing in AutoCAD and was followed by the production of a 0.3 scale half-model – shades of boat building. After sanding this so it looked right the laminated sections were each scanned on a flat-bed scanner, and then inserted into the CAD drawing to allow tweaking of the shape of the templates.
Then construction for real started, with endless hours of cutting plywood just with a circular saw and, mainly, a jigsaw: with 30 pieces of plywood per speaker I was soon cursing the frugal approach that precluded handing it all over to a CNC-equipped workshop (partly due to lockdown-derived impoverishment and partly due to the fact that I wanted to build them myself). Then gluing the plywood sections together to make half-speakers: it sounds easy, but the awkward shape and slippery glue were a difficult combination, resolved by adding a few laminations at a time and holding things in place with piles of books. Fiddlier still was making the front plywood baffles for mounting the drivers: the drivers sit in rubber O-rings with fixings via threaded bolts out of the back of the speakers, so millimetre precision was needed (achieved by spinning my router round a nail). Then back to hard graft: heavy sanding – mainly with a belt sander – to get into something near the final shape. I must have clocked up more hours with a mask on in my own garden than anyone else in lockdown. With initial sanding done it was on to stuffing the lower half of the tapered tube with woollen jumpers (almost solid at the bottom), add cables, and then, without stray bits of wool mucking things up, glue the two halves together: more wool and then – for the top part – speaker wadding was added through the bass driver hole.
Setting out paper templates, and working how to fit to the 8ft x 4ft plywood sheetsPlywood laminations cut for one half-speakerThe not-so-sophisticated gluing process: lashings of Titebond and precariously-balanced books.The first half-speaker nearly finished, with drivers loaded. The tweeter has its own (plastic) exponential tube fitted, and both drivers are held in place by bolts through the back of the cabinet – requiring careful alignment and rubber fittings to avoid resonance issues. The interior was intentionally left with a stepped effect since sanding (which would have been difficult) would not have improved the performance.
With one speaker roughly made, it was time for loading the drivers for testing. With no access to a handy anechoic chamber this time, the garden on a windless day was the next best thing – getting away from any nasty indoor reflections. The neighbours must have thought it odd as I fed sine-wave sweeps through the speakers: aliens in the village? Measurement (bass driver and tweeter separately) was via a Dayton Audio EMM-6 measurement mic, using its individual calibration data and REW software (useful not just for measuring room EQ). The results were looking hopeful, so data was sent to the crossover designer (a renowned loudspeaker designer, so in safe hands), then the crossovers were assembled for me (using hand-wound inductors, made on a slowly spinning lathe), and the tests run again, this time with both drivers in unison: superb results, with impressive waterfall plots revealing negligible resonances and reverberation. During the to-ing and fro-ing of crossover design and build, I ground on with construction of speaker no. 2, and then both speakers were given the finer finishing sanding, crossovers were fixed, and then the speakers were treated to several coats of high-gloss yacht varnish.
Gluing the two half-speakers together: not the easiest shape…Measuring a speaker in the garden: the poor man’s anechoic chamber.Almost there: soldering connectors to one of the varnished speakers.
And the sound? After teething troubles with one bass driver (an elusive intermittent distortion caused by a loose connector in the driver – as I said, these were spares, and had been knocking around on a shelf for a few years) were sorted, absolutely superb! REW results indoors were exactly as anticipated and aimed for. But, more to the point, listening reveals the sound to be uncoloured, responsive, clear, with a good stereo image, and easily sufficient bass (something I was wondering about) for the size of room. At some points the complexity of utilizing the new drivers seemed daunting (compared to the old B&W ones), but, in the end, it was worth it: not only was there an aesthetic upgrade (the original plan), but an equal acoustic one. For the cost of several sheets of ply, a lot of glue and jigsaw blades, some crossover components, a tin of varnish and countless hours (which were going free during lockdown anyway), I’ve ended up with some unique-looking speakers that to better performance-wise would cost an arm and a leg. And such a satisfying feeling. Given that the mark 1 snakes served over 25 years, these ones will see me out.
Ah, I forgot to mention, though, that there is one drawback: this time round you can’t balance a mug of tea on the top of the speakers…
Rob Moore has many talents – artistic, musical and practical. Over the last three years I’ve enjoyed making short and simple videos that showcase some of his amazing rolling-ball sculptures: that is, sightly whimsical, but endlessly fascinating, kinetic sculptures drawn in stainless-steel wire and given motion by balls raised mechanically and then descending under gravity. Rob’s most ambitious creation to date has been his ‘Kinetic Lungs’, but, in early 2021, he is just about to start on a far more complex sculpture still: the Brain.
There is a small number of like-minded individuals making these kinetic sculptures, spread across the world and linked mostly by the internet. Many of the videos posted showing these sculptures have soundtracks dominated by music, but, to me, the noise of a rolling-ball sculpture is a significant part of its character: the background rumble of the mechanism that lifts the balls up; the swish of the balls as they descend the tracks; the clunks and thuds as gates and switches move; and the frantic spinning of balls in cones. So my approach to the films has been to focus on the sound, tracking, as it were, an individual ball through the sculpture, in a slightly hyper-real way. In each case I recorded the mechanical lift system separately, then turned this off and recorded the progress of an individual ball close-up, section by section with a spaced pair of mics giving a rather exaggerated stereo sound, as if leaning in close to the sculpture. This was then pieced together, overlain with the mechanical rumbling track (with this turned down, to match more closely how you perceive the noise in the flesh) and (and this is the complex bit) multiplied and synchronized with the multiple balls in the video. The video itself in each case has the whole sculpture shown as a static reference shot, with a split-screen showing changing close-ups, but all synchronized: a complexity of audio and visual synchronization that gives a very simple-looking end result, which lets Rob’s rolling-ball sculptures do the talking.
With the upcoming Brain the filming requirements will be much more complex, with pieces to camera by Rob and behind the scenes filming of its making. I’ll post more about the filming and sound-recording aspects of what will be a long build: and Rob will doubtless post updates on his progress on his Facebook page.
