Binaural Recording, Pt 3

After my second experiment in binaural recording, described in Part 2 of this series, the third area of my research focused on the importance of the ears, rather than the head, to the quality of the recorded sound.  The pdf article I had orginally read on the subject had stressed the importance of the shape of ears, rather than the presence of the head, and at least one range of well-known commercial binaural microphones, the 3dio, consists of ears without a head.

So, I set about sourcing a pair of realistic ears.  Some of these can be prohibitively expensive, but I found a type that were more reasonably-priced (a little under £6 for the pair, from China), moulded in silicone, and intended for acupuncture, it said, or medical study.

When they arrived I was surprised, as they were heavier, softer and more flexible than I was expecting – nothing like the hard, fixed ears of the dummy head.  Much more realistic, and softer even than real human ears.

Unfortunately, as you can see here, the right one – on the left in the picture – had been rather crushed in the post, with the ear lobe bent right over:

The insides of the plastic bags seemed quite moist, so I was hoping that if I pulled the ear lobe back and stored the ear upside down, with a bit of weight on it, it would still be malleable enough to return to something like its original shape.  Perhaps when taken out of the bags, the ears would then harden up in the right shape.

In the  end I had to use a dab of hot glue to pin the ear back to its proper place.  The ears became less moist, but not really less floppy after being out of their bags for a few days.

*

The next part of the plan was a pair of ear muffs, readily available from Chinese sources on eBay at about £1.  They come in all sorts of colours and finishes, but the ones I picked on had a fur interior and a faux-leather finish on the outside:

The reason I chose this type was that the fur would provide a certain amount of sound deadening, and the faux-leather would take hot glue.

More hot glue?  The glue, of course, was needed to attach the silicone ears to the ear muffs!

The purpose of this was to be able to use the ears in different contexts, either attached to a dummy head, or used on their own, relying on the ear-shape for the binaural effect, rather than the whole head.

Accordingly, I first obtained a cheap polystyrene dummy head and inserted some long nails to act as a support for the ear muffs:

It took a few tries to get the exact locations for the nails, but I fairly soon had them in the right place to hold the ears at the correct angle:

I then cut a small block of polystyrene to a suitable size to fit the ears on as a free-standing unit.  I estimated that 14-15cm would be a suitable separation distance for the two ears and trimmed the polystyrene accordingly:

The final and most important stage was to add the electret capsules inside the ears and attach a twin phono lead for the output to the preamplifier.

I used smaller capsules than the ones I had used in the dummy head, as it was difficult to make holes in the silicon ears.  In the end I used my polystyrene cutter, which has a small heated blade, inserted the cable between the silicon block and the ear muff backing, attached the capsules and then pushed them down into the ear canal towards where the ear drum would be.

These photographs illustrate this process:

You can also see a certain amount of melted silicone debris, which I had to remove.

After tidying this up and testing the sound was OK, I superglued the cables along the back of the ear muffs to keep the microphones from moving, and the project was finished.  As soon as there is some better weather, I’ll be able to go out and test the three different binaural systems together.

Ultrasonic Field Recording

I had been looking for ways of recording sound from scenes which was not immediately obvious.  I had developed an inductor-based recording device which could pick up electrical noise, and the next area I wanted to explore was ultrasonic sound – that is, sounds which are too high for our ears to hear.

A recording of sounds which are too high for our ears to hear, of course, would not be too interesting – you wouldn’t be able to hear anything on the recording, either; but I got the idea of what to do with them from a type of device available commercially in finished or even kit form – a Bat Detector.

Bats make noises which are mostly ultrasonic – too high for our ears to hear – when they hunt and communicate.  What the classic bat detector does is pick up these ultrasonic sounds, amplify them and lower their pitch by an octave or two, so we can hear them.

But not only bats make noises in the ultrasonic region, many noises around us contain ultrasonic elements which we can’t hear as well as elements which we can hear.  Lowering the pitch of the ultrasonic elements would enable us to appreciate the fullness of sounds which at the moment we only hear part of.

There are several ways of changing the pitch of ultrasonic sounds, and there are many circuits available on the internet utilising these methods for producing a bat detector.  The three main ways are:

1. Frequency Division, in which the very high-pitched bat sounds are converted into square waves which can be digitally divided by – typically – 10, to produce a much lower sound, within the range of human hearing. For example, a bat making calls at 50kHz, when divided by 10, will sound at 5kHz – quite high, but well within our normal hearing range of 20Kz to 20kHz.

2. Time Expansion, where sounds are recorded digitally in the bat detector at a high sampling rate, then played back at a slower rate.  This is a common method of pitch-changing in the world of digital sampling (and the method I used in the computer-based Black Widow sample manipulator).

3. Heterodyne, which works on the same principle as the electronic musical instrument, the Theremin: the high-pitched sounds are mixed with equally high-pitched sounds produced by an oscillator inside the bat detector. The aim is to tune the bat detector to produce a slightly different pitch to the bat, as the output is designed to be at a frequency which is the difference between the two pitches. In this case, if the bat is making calls at 50kHz, for example, and the bat detector is tuned to 45Hz, sounds will again be heard at 5kHz.

This latter turned out to be the method I used – in fact, I bought a bat detector kit.  The best commercially available bat detectors are very expensive, providing a wide frequency range to work within, and often providing more than one way to hear the bats.  I should make it clear at this point that I wasn’t particularly focusing on bats – I wanted to listen to any ultrasonic sounds that might be in the vicinity; but I wanted as wide a variety of sounds as possible to be detected and brought into hearing range.

So I found a neat-looking and very reasonably-priced kit, the Franzis Bat Detector, which was available from a number of sources, all at around £20-£25.  It comes in an attractive and quite sturdy cardboard box, which can serve as the container for the circuit when made up, with art work for the two potentiometers required – for frequency and volume – and holes behind which the speaker is attached.

Inside the box is a very small PCB, no more than 3″ long, already populated with about 25 tiny SMD (Surface Mount Device) components:

Together with the board is a plastic bag containing a handful of non-SMD components, which you solder in place yourself.  These include a battery clip (not shown), a 5v voltage regulator, a few capacitors, an LM386 i.c. amplifier, the two potentiometers and an ultrasonic receiver – in appearance rather like a large electret microphone capsule.

Together with the components, there was a nicely-produced and informative booklet, containing general information about bat detection, an explanation of how the circuit works, a circuit diagram and comprehensive instructions for assembly and testing.

*

Beginning at the front end of the circuit, there are many different types of ultrasonic receiver available, and the majority are rather expensive.  The one that came with the kit is the most popular of the more reasonably-priced ones, but is designed to work best at 40kHz, with quite a narrow band of frequencies in which it works at its greatest efficiency.  This is because it is designed to work in precisely that way, usually being paired with an identical-looking 40kHz ultrasonic transmitter, and commonly used together in detection or distance measuring applications.  I was concerned that this frequency restriction might limit the sounds the detector was able to pick up, bats or otherwise, but there is a certain amount of pickup outside the intended operating range.

*

This particular circuit would probably not suit it, but a possible alternative to the ultrasonic receiver in some circumstances would – surprisingly – be an electret capsule.  Although these are sold for use in ordinary microphones, some have good ultrasonic capabilities.  It’s hard to know which ones, and to what extent they might be useful in this regard, as figures are not normally released for frequencies above 20kHz, the normal extent of human hearing.

However, one capsule known to have a good response even in the environs of 100kHz is the Panasonic WM-61A, one which has been tested and used in this way.  Unfortunately, this particular capsule was discontinued more than a decade ago, and remaining ones are getting more and more expensive, even if they can be found.  Some are still advertised on, for example eBay, but I was put off by warnings of possible fakes, whose frequency response would not necessarily be the same.

A good currently available alternative is the Primo EM258 from FEL Communications.  At over £5 each, these were 20 times as expensive as the unbranded electret capsules I’d bought before for other more conventional microphone projects, but they are not much more expensive than genuine Panasonic WM-61As these days, and have been tested and shown to have a good ultrasonic response (better, in fact than the Panasonics, according to FEL).

JLI Electronics manufacture the JLI-61A, which is intended as a direct replacement for the WM-61A, but it wasn’t clear that this was available at a reasonable price in the UK.  In the US this would be a good  alternative, at half the price of the Primo.

FEL, incidentally, also advertise a potentially excellent alternative, a tiny SMD-style MEMS microphone.  Normally these things are practically invisible to the naked eye, but FEL have installed one on a breakout board like this:

The microphone itself is a Knowles SPU0410LR5H-QB, as the legend on the PCB suggests, with a sensitivity to ultrasonic frequencies up to 200kHz and beyond.  It was almost twice as expensive as the Primo electret, but would, no doubt work very well, and that price, just under £10, is not at all unreasonable compared to current good quality alternatives.

Alternatively, the cheapest way to obtain a second ultrasonic detector – other than ordering direct from China – might be to purchase a module like this:

Its purpose is distance measurement – the device on the left, marked ‘T’, is an ultrasonic transmitter, the one on the right, marked ‘R’ is a receiver.  It would be a few moments work to detach the receiver from the board, and attach it at the beginning of the circuit.

*

As for the construction, I planned to fit the circuit inside one of the small boxes I had previously used for microphone preamps; so I connected the small PCB to sockets in the box for power and audio out – I intended to use headphones instead of the speaker supplied.  I also added an extra socket for a line out from the wiper of the volume control.  Together with an appropriate preamp (for example, the one I use for my contact mics), this would enable me to record the ultrasonic sounds I was picking up.

 

I attached the ultrasonic detector to the front of the box with hot glue, and attached a pair of 2.5M bolts for the two aerials, which had threaded bases (in the end I only used one aerial).  I also added an extra socket for attaching an external aerial or detector; a plug here would disconnect the internal ones.

The remaining parts of the circuit were: a buffer/amplifier for the ultrasonic detector; a high-frequency oscillator, based on 555 integrated circuit; a mixing/heterodyne circuit, and an audio amplifier, based on an LM386 (the only part of the circuit which uses the full 9v available from the PP3 battery).

*

The difficulties with the first part of the circuit  – the buffer/amplifier – if you were to build it yourself, are to do with size.  The specified transistor for this amplifier, the BC849C, is a tiny, tiny 3-pin SMD device.  I didn’t have a 5p handy – physically the smallest coin currently in use – but I did have a 1p, which is only a little bigger, and a BC849C, and this is how they compared:

It would be quite a task trying to attach wires to this miniscule component – but at least they’ve separated the 3 pins onto separate sides.  I obtained this one just as an example – the actual one used in this circuit was already happily soldered to the PCB by the makers!

*

The mixer section is based on a CD2003 chip, ‘originally developed for radio receivers’, as the kit booklet says, ‘the core of an AM/FM radio with oscillators, mixer stages, intermediate frequency amplifiers and demodulators for the two ranges’.  In this design, only the AM preamp and AM mixer stage are used.  According to the booklet, the i.c. ‘offers a total amplification of 40 dB and a suppression of the input signal of -20 dB. The output of the mixer provides a low pass filter for an additional damping of the input signal.’ – a very handy chip for the task in hand.  It is possible to get hold of these, but they are not nowadays common or cheap – except from China, it appeared.

*

After connecting everything together, I plugged the headphones in and tested the detector out, using the preamp I had constructed for the piezo contact mics.

The device worked well: I detected ultrasonic sounds from jangling a bunch of keys and from rubbing my finger and thumb together – everyday sounds known to have a significant ultrasonic component – and outside in the evening I was pretty sure I detected some genuine bats.

This view of the front of the device shows the two potentiometers which need to be accessed when the device is in use: the tuning control on the left, which adjects the frequency of the internal oscillator and effectively ‘tunes in’ the high frequency noise picked up by the ultrasonic detector; and on the right the volume control.

This view of the back shows, on the left, the unit with the 9v battery attached (with velcro, in my usual way); and on the right, without the battery, but showing the caption for the switch, indicating that either the ultrasonic ‘mic’ or the attached aerial can be selected; a 3.5mm plug in the ‘EXT AERIAL’ socket on the front disconnects the switch, so a different size or kind of aerial can be used.

The following sound file gives examples of some quick recordings I made with the ultrasonic detector.  Unfortunately, it’s now late November – much later than when I first tested the circuit out – so no bats flying around.  Instead, I just went round the house for 10 minutes, picking out a few promising locations.

So, you can hear fingers rubbing together, the laptop, TV set, some unexplained radio-tuning type sounds, a low-energy light-bulb, jingling keys, and water streaming slowly into the sink from a tap. The laptop, TV set, radio-tuning and low-energy light-bulb were recorded with the aerial, the others with the ultrasonic microphone-type detector.

The keys are particularly interesting, and I look forward to trying this on some natural sounds outside – especially bats when they emerge from hibernation next spring.

Making music with the BBC Micro:bit, Pt 1

Recently I was lucky enough to be given a BBC Micro:bit as a present.

What is a Micro:bit? – It’s a micro-sized computer, half the size of a credit card, with a row of input and output pins along one edge, Bluetooth capability, a display consisting of 25 LEDs in a a 5×5 pattern, and able to be powered and programmed via USB.  For ultimate simplicity, it can be programmed directly from the website www.microbit.org, via WebUSB using the Chrome web browser.

When it was introduced, its purpose was to ‘encourage children to get actively involved in writing software for computers and building new things‘, rather than being merely consumers of media, and around a million of them were given away to schoolchildren by the BBC.

Once I found out that it was considered suitable for primary school children, I was greatly encouraged that I would be able to write programs for it; and, in fact, there is a particularly straightforward way of doing this (as well as Javascript and a version of the Python programming language): that is, the Microsoft MakeCode system, which involves arranging and sequencing interlocking ‘blocks’.  You still have to have the programmer’s mentality, but without having to remember the precise wording and syntax.

As shown in the pictures above, The Micro:bit has a row of about 20 pins on its bottom edge, which have different digital or analog functions, and 5 of them, in particular, are easily accessed with crocodile clips or 4mm banana plugs.  These are +3v (actually 3.3v) and Ground, and analog pins P0, P1 and P2.  As a matter of fact, these last 3 pins can be operated simply by touching them!  You may have to touch Ground with your other hand to make sure they work, though.

Two tactile switch buttons are also available for incorporation into simple programs, and there is a tactile ‘Reset’ button on the back, which causes the Micro:bit to reboot.

For more complex applications – which can even include operating external devices such as servos – a range of edge connectors is available, which can aid permanent connections or take advantage of the smaller input/output pins.

There seemed to be several different ways in which I could make use of the Micro:bit’s capabilities.  First of all, it can, with a speaker attached directly to one of its analogue pins (typically pin P0), make monophonic square-wave sounds of its own, including playing a number of tunes stored in its own memory; secondly, it can be programmed to output MIDI messages; and thirdly, it can utilise and output data from its onboard detectors, which include light and temperature sensing, a three-dimensional accelerometer, and a compass/magnetic sensor.

I decided to start by using the Micro:bit’s internal sound-generating capabilities.

For this, I decided to make a module into which the Micro:bit could be inserted, and which would contain a power supply, speaker and various input and output sockets – all the hardware needed for a free-standing Micro:bit-controlled musical instrument.  That is the subject of this post.

*

I began with an empty Stylophone case, left over from an earlier project.  I left the speaker in, and the two slide switches on the front left.

In order to incorporate the Micro:bit, I bought a ‘proto-board’ – an edge connector into which the Micro:bit slots at a right-angle, with a section of stripboard attached for ancillary circuits to be built on.  I chose the right-angle one because I needed the Micro:bit to be visible when in use, and to have access to the front and back of the Micro:bit where its sensors are located.

I used this board for a couple of things: a simple op-amp based buffer/voltage follower, and an audio output transformer to be used with an alternative to the speaker, a piezo element.

I also bought a sturdy protective cover for the Micro:bit to ensure that it suffered no damage while being plugged in and out.  This particular type of cover leaves the edge with the pins exposed for just this sort of application.

This picture shows the board in situ, kept in place with matchsticks used as locating pins.  The transformer and op amp are on the right-hand side; on the left are a 4-pin i.c. socket and capacitor left over from an experiment with an audio amplifier, which I decided not to keep.

The voltage follower is the simplest possible op-amp circuit:

I used one half of an NE5532, as that’s what I happened to have available, but any op-amp can perform this task of ensuring the output has a suitably low impedence level.

I’m not sure how useful the transformer is in this exact context, but I learned of its use with piezos from this Nic Collins video on YouTube and had bought a handful of them for this kind of use.

These pictures show how the piezo was fixed into the former Stylophone case, from the inside (L) and outside (R) in the place previously occupied by the tuning knob, which protruded from the bottom of the case.

The Micro:bit can also be powered through thie proto board: mine had a terminal block already installed in it for +3v and Ground; I powered the unit with a PP3 battery, as I am wont to do, so added a small adjustable voltage regulator before the terminal block, set at 3.3v.  This meant that the Micro:bit was receiving its correct voltage, but more volts (up to 9v) would be available for any additional circuitry required.  Two LEDs show whether the 9v and 3.3v parts of the circuit are powered up.

While working on the inputs and outputs, I added the following: an external 9v power in socket; send and return sockets so that external tone controls (such as these) could be used; sockets for an external speaker or headphones and an external piezo; a line out socket for easier mixing or recording; and two sockets connected to analog input pins P1 and P2.

Pin P1 was connected to a 10k lin potentiometer, one end of which was connected to 3.3v, the supply voltage for this part of the circuit, the other to 0v; in this way a continuously variable voltage could be available at pin P1 for control purposes: the Micro:bit would interpret this as a number between 0 and 1023.  If something was plugged into the P1 socket, the potentiometer would be disconnected and the external signal read by the pin instead.

Pin P2 was not connected, unless something was plugged into the P2 socket.  A second 10k lin potentiometer, set up exactly like the one for P1, was added, terminating in a 3.5mm mono plug.  This was available to be plugged into the socket and used for the same purpose if a program were to require it.

In use, I found I had problems with the circuit if the PP3 battery wasn’t new or, if rechargeable, well charged; fortunately, with the external socket it would be possible to use a mains adapter.

*

I set the mIcro:bit up to make some trial tones, but the sound was rather quiet, through both speakers and piezo, so I added a small amplifier.  This was a very economical (99p) module from a UK eBay seller, based on the PAM8403 chip.  This is a 3W stereo amplifier chip, but I would only be needing one of the channels.

It required a 5v supply, so I powered it from the 9v source and added a small 7805 5v regulator.

I put a 10k log volume control before the amplifier input, with the line out socket between the volume control and the amplifier input.  If a plug is inserted in the line out socket, the signal to the amplifier is cut.

After the amplifier there is a 3-way switch to choose between internal speaker, external speaker or headphones, and piezo element.

The following picture shows the majority of the interior at the end of phase 1 of construction (but before I added the two LEDs):

Features marked: 1 = the original Stylophone switch PCB, repurposed for 9v on/off and 3v on/off; 2 = volume control; 3 = the original Stylophone speaker; 4 = the 5v voltage regulator and amplifier PCB; 5 = the 3.3v voltage regulator PCB; and 6 = the 3-way audio output switch (internal speaker/external speaker or headphones/piezo element).

On the exterior I added three tactile switches to replace the two switches A and B on the front of the Micro:bit.  The first two switches simply duplicate the effect of the two switches on the Micro:bit; with the aid of a couple of diodes and a 4066 digital switch i.c., the third tactile duplicates the effect of pressing A and B at the same time, an action recognised by the Micro:bit.

I didn’t add any pull-up resistors on the outputs of the 4066 as the Micro:bit already has these internally.

*

After this I had a working module, so it was time for some programming!

*

I started with a couple of simple programs to make use of the light sensitivity and magnetometer functions.

While experimenting, I connected the Micro:bit to my Macbook with a USB cable, opened the Chrome web browser and wrote the programs using MakeCode on the microbit.org website.

The following picture shows the main features of the coding page:

1 = the simulator.  Most – but not always all – of the coding you write is automatically simulated here.  As you can see, the first instruction of my program, to display the number 1, is shown, as is the indication that I have chosen instructions which require the speaker to be attached;

2 = the button you press to load the program into your USB-connected Micro:bit.  WebUSB has not yet been implemented in most web browsers, which is why I use Chrome to do this.  Incidentally, if you’ve made any syntactical errors in the program, you will be told when you click this button;

3 = the area to name and save your program as a hex file.  The alternative way to program the Micro:bit is to open its window and drag and drop a hex file onto it;

4 = where you choose to program using the blocks, as I have done, or to write the code in Javascript;

5 = the area where the program is written.

*

As for my first program: the light control on the Micro:bit takes advantage of the fact that if the current to an LED is inverted, it becomes (slightly) sensitive to light.   The Micro:bit has an array of 25 LEDs on it, and it’s possible to use 9 of these together to record light levels on and around the device.

I found that the most useful way to use this feature was to set the Micro:bit up to sense the ambient light level, and then react to brighter lights being shone onto the LEDs, rather than the other way round (i.e. reacting to being shaded from the ambient light).

In this case, the reaction I programmed was to sound a note, then increase the pitch of that note as the light level increased.  Initially I just used the torch in my mobile phone, but I had already made 2 devices which could be used to control the Micro:bit in Light Reactive mode, the UFO and the Shuttlecraft.

As I said, the Micro:bit was programmed to start up in a waiting state and display the number ‘1’:

Then, when its Button ‘B’ was pressed – or the Button ‘B’ on the top of my module – it would run the part of the program related to the Light Reactive Instrument and display the number ‘3’:

The Micro:bit outputs a number between 0 and 255 to represent how bright the light level is.  0=dark and 255=bright.  The variable lightlevel in the program is set to be 20 times this number, i.e. between 0 and about 5,000.  The opening pitch chosen, 220Hz (the note A below middle C) and the range of values represented by lightlevel is designed to allow the Micro:bit to output pitches between 220Hz – a low to medium note – and about 5kHz – a very high note.

The variable lightambient is used to sample and remember the normal light level around the instrument on startup.  If the light level is no greater than this, the instrument makes no sound, the idea being that it should remain silent until a light is deliberately shone on it.

*

The second simple instrument I programmed was, in essence, the same thing, but reacting in this case to the presence of a magnet.

The Micro:bit measures magnetic force, as it says above, in microteslas (µT).  It turned out that the range of readings it gave did not need scaling or multiplying like the light level readings did, but produced a useful range of pitches without any changes.  So in this case the variable magforce was equal to the microtesla reading and gave rise to an output at the same Hz as the reading.

In this case, a magnet in the vicinity of the Micro:bit caused a pleasing arpeggiated effect, increasing in pitch the closer the magnet came to the Micro:bit, and decreasing as it was moved further away.

Again, the variable magambient was inserted in order to stop the instrument from sounding until the activating magnet was intentionally brought close to it.

I bought a neodymium (NdFeB) magnet especially for this purpose, as these are particularly powerful – up to 20 times as powerful as conventional ferrite (iron) magnets, in fact.

Neodymium magnets are graded from 28-52 according to their strength.  This one was an N52 (highest power) type.  It works very effectively, but has to be kept well away from any metal parts of the instrument – indeed metal parts of anything – as it will stick very easily and very strongly to them.  I wasn’t sure, but it even seemed to be affecting the speaker in some instances – a speaker, of couse, being driven by a magnet.

The only negative thing about this second instrument is that every time you reboot the Micro:bit, the compass/magnetometer has to be reset if you come to use it.  This is not, unfortunately, a simple matter, as you have to turn the Micro:bit every which way, with your progress shown by the LEDs lighting up one by one.  Only when all 25 of them are lit can you proceed.  This is obviously critical, if you genuinely want an accurate compass direction, but not so critical if all you want to do is make entertaining noises . . .

*

As I mentioned earlier, I had added two potentiometers connected to analog input pins P1 and P2.   I didn’t have an immediate use for these, but I thought it would be handy to have two variable inputs – for tuning or transposition, for example.

In preparation for Phase 2 of the instrument, I made use of the input to pin P1, and programmed it in the following way:

At the beginning of the program the potential input from pin 1, 0 – 1023, is divided into 21 equal sections, and scaled down to 0 – 20.  Each of the 21 divisions is numbered. ‘notenumber 0’, ‘notenumber’ 1′, ‘notenumber 2’, etc.

There are 20 notes on the former Stylophone keyboard, so there is one number for ‘off’, and one number for each note on the keyboard, ranging from 220Hz (‘Low A’) up to 659Hz (‘High E’).  When the potentiometer connected to pin 1 is turned fully anti-clockwise, the instrument is silent; as it is slowly turned clockwise, each note is stepped through in turn, until the highest note, ‘High E’ is reached.

This experiment proved that it should be possible to connect the keyboard to pin 1 and use the Micro:bit like a Stylophone.  This what I intend to do in Part 2 of this series.

Meanwhile, here are some recordings of these two instruments.  There are 8 short recordings, following this pattern: first the Light Reactive instrument through the internal speaker, external speaker, internal piezo element amd external piezo element; then the Magnet Reactive instrument through the same four media.

The recordings with the external piezo, glued to the bottom of the tin, are particularly interesting.  My experiments with piezos (beginning here), have so far only been in using them as microphonic elements, in particular as contact microphones, and as part of custom-made percussion instruments, and although the introductory article refers to their use as speakers, this is the first time I had actually used one that way.  Hopefully, there will be an opportunity to complete my survey of piezos by looking into this in more detail.

The tin, in fact, had been prepared for use as a kind of drum – and could easily function that way if attached to a suitable preamp.  It is quite large in size – 22cm (about 8.5″) in diameter and 14cm (about 5.5″) deep – and would make a very effective drum.  In this instance, using it as a speaker adds a noticeable reverb effect to the sound of the Micro:bit instrument.