Choosing an Audio Recorder
Faced with an endless and sometimes confusing range of covert audio recorders and intercept devices, how do you choose?
At Somerdata, we have selected for you the pick of the best around, and where we don’t carry them ourselves, we will recommend the best solution for you.
Unlike Video recording, audio can be quite difficult to understand and there are still a plethora of formats, rates, technologies and so on to choose from.
So, first try and define what you want to achieve, then compromise to what you can get ( and afford!).
Here are some questions to pose, only you can supply the answers.
1. Is my target benign or hostile? How will they react to discovery of their being recorded? Do I need to tell them beforehand? ( you might need to do some research on lawful intercept and fair usage in the territory in which you operate.)
2. Is my target stationary or mobile? Do they visit one location or many?
3. How close can I get?
4. Can I get to the target site(s) safely before and /or after the surveillance period?
5. How long is the surveillance period? Is surveillance unattended? Does it need immediate reaction to events?
6. Do my recordings need to be presented as reviewable evidence, especially in a court of law.?
7. Does my device need to be self-powered? How physically robust does it need to be? Are there size and packaging constraints?
8. Does the device need to be protected from counter-measures and interception? what happens if recordings fall into the wrong hands or transmissions are intercepted by a third party?
9. How much is my data worth? How much am I prepared to spend to collect it?
If you’ve answered ALL of those questions then you probably already know enough to make your own selection and you don’t need any more advice. Carry on!
For everyone else, let’s look at the options, technologies and practical solutions.
Firstly, if you carry a mobile phone, then you already have a mobile audio recorder. Put it on the desk, point it at your target and press record. The quality will be ok and you’ll get something. Now here’s the weird thing. When you come to play it back, you’ll suddenly notice all sorts of strange noises you didn’t hear when you made the recording. Rustling paper, room echo, road noise, and a hundred other liltte noises, some quite distracting. With luck you may even have recorded something useful. Why is this?
Well, the human auditory system is a remarkable instrument. It can filter out extraneous noise, correlate muti-path echoes, focus on point sources and acts as a very sophisticated filtering system especially when coupled with eyesight. Human ears have a very wide dynamic range (>120dB) meaning we can hear whispers and jet engines with equal clarity, but we also non-linearly favour speech frequencies – we hear in the range 300 to 6 kHz more distinctly than in frequencies outside this range. And this also varies with age.
Crucially though, we hear in Stereo using two ears. This not only gives us twice the sensitivity but we are able to give spatial resolution to what we hear – where it is, how fast it’s moving, separation of multiple sources into distinct focal points.
So technically, the mobile phone is going to struggle a bit to compete. It’s probably mono, limited dynamic range, limited frequency range and doesn’t match the human ear response.
Which leads us to the grown-ups – dedicated recorders and transmission systems designed specifically to work in this environment.
Dedicated audio recorders for use in covert and /or surveillance operations generally need to be physically small, robust, high quality and easy to operate. In addition, they may need to have long recording time, be secure if mislaid, and invisible to counter-measures. Some if not all of these are contradictory and require practical compromises.
Lets look at some of these.
Quality – by which we mean intelligibility and repeatability.
We need to be able to understand the recording and probably listen to it many times without degradation. Repeatability comes from digitising the audio, then using a high-quality medium to store the resultant digits. Nowadays this means solid-state (semiconductor) . Not infallible, but better than mechanical systems like tape, disk or cd/dvd which are easliy damaged.
Intelligibilty comes about through a combination of dynamic range ( the ratio of the loudest sound to the quietest) and frequency range ( bass to treble).
Digitisation involves cutting your nice analogue signal into thousands of bits that take up a lot more space and take longer to transmit than the original, so the concept of compression was introduced to compenaste. This is not the place to debate compression algorthms, suffice it to say that compression, especially non-reversible, is generally bad for intelligibility. Digitisation also creates its own limits. As a rule of thumb, you need to digitise at twice the highest frequency you want to record at – as a practical example, 8kHz of voice bandwidth needs to be digitised at 16KHz. ( For comparison, audio CD quality digitises at about 44kHz to get a HiFi acceptable bandwidth of 20kHz). Note though that the number of bits created also doubles, so does the storage capacity and the transmission rate. In general you need the highest sampling rate you can afford ( although there is no point in outsripping the capability of you microphone). And from an intelligence gathering point of view, you may want to include a wider bandwidth than is strictly necessary for voice, to improve spatial information or pick up extraneous sound information that gives context to the operation.
The other determinant of quality is the number of bits per sample, the quantization level. 8, 10, 12,14 and 16 bits are common. The practical effect of quantization is to largely determine the noise floor, or put another way, the overall signal-to-noise ratio and thus the dynamic range. The more bits per sample used, the lower the noise floor and the wider the dynamic range; so 10 is good, 16 is better. Translated to analogue quantities, 10 bits is about 60dB , 16 bits can represent about 96dB of Signal to Noise ratio. Of course this is not the only determinant of system noise - power supplies, microphone and pre-amplifier semiconductor noise may combine to dominate. And as usual there is a price to pay for more bits, in power consumption, download/transmission time and storage space required
Discrete – or in other words very small and unobtrusive, preferably invisible. Packing a lot of data into a very small space is a challenge for both microelectronics and power supplies. In fact, portable device size is almost always determined by the battery size which in turn is determined by the power consumption and operating time of the electronics -which in turn is determined by sampling rate and storage technology…… etc.!
Invisible is still a bit tricky but consider hiding in plain view – a pen, a key fob, a usb stick or a bluetooth earpiece are all available and ignored by most people.
Secure – faced with the inevitability of losing your recorder at some point, what about its content? Would you want that falling into the wrong hands ? (ie not yours). Securing the contents of the recorder is vital, so consider using a record-only device. This may sound strange but what it really means is being able to retrieve the data only under controlled conditions. A password protected recorder/replayer is a good start but most passwords can be cracked quite readily. They are created by people, who like to be able to remember them, so passwords tend to submit to computerised brute force attack quite readily.
Encrypting the stored data is another help, now you need a password ( you did set one didn’t you….!) and a key. But encryption also takes up storage space, consumes power and is not readily available on simple devices. Storing the data in an unfamiliar format, proprietary or obscure is another method. users now need access to a format converter which may be unpublished or simply unavailable. This is almost as effective as encryption and effectively limits access at no overhead cost to the device.
Alternatives to Recorders
The obvious alternative to recorders is direct transmission and the uninitiated often say, “all i need is a microphone with a transmitter and i can use wireless – i don’t need a recorder near my target”. Let’s examine that in the same technical way as we did above.
Wireless transmitters suffer many of the same constraints as recorders with a few added. Firstly, a microphone or audio gathering system still needs to be in the right place. It needs a power supply, a transmitter and an antenna. If it needs to be externally controlled it needs a receiver as well. Wireless transmissions are subject to strict regulation – you can’t just use any part of the wireless spectrum for your transceiver. In fact there are very few parts of the spectrum available to non-broadcast or military devices and they come in distinct bands, for ecample 900MHz, 1.8GHz, 2.5Ghz, 433MHz, 27Mhz and so on. Why is this important? Firstly, these bands are already heavily used -cellular phone networks, bluetooth, zigbee, low power radio, radio controlled models and so on. Secondly and more importantly, their transmissibility varies with frequency or wavelength if you prefer. Short wavelengths ( higher frequencies) travel in straighter paths, the so-called line of sight problem. Short wavelenght radio doesn’t travel round corners. So shaped antennae are required that ‘point’ signals in the desired direction. Perversely, for this argment, the more useful longer wavelengths that can travel further and bend a bit, need physically larger antennae. In the case of the longer wavelengths, a quarter wave antenna can be half a metre long. Go lower in frequency and we start to talk several metres (or several equivalent metres) with folded arrays and reflectors etc. These can hardly be described as discrete.
So surveillance trnasmitters of this type tend to concentrate on the shorter wavelengths. For example Bluetooth operates in the 2.5GHz region. Consequently the range is limited and if someone happens to park a bus or a truck in front your transmitter, or it rains heavily, or there are a lot of other bluetooth devices around then the range may be zero as far as you are concerned!.
The effects of this can be overcome a little by boosting the power of the transmitter, bt leaving aside the increased power supply or batter needed to do this, this leads to the next problem -detection and counter-measures.
Clearly, a wireless transmitter by design is capable of being found by a receiver. If it can be found by your receiver then it can be found by someone else’s. Or more likely by a sweeping spectrum analyser designed to find it. Strategies to overcome this include spread-spectrum, chirp or burst transmission and sweep detector detection. These add to the complexity, cost and detract from the reliability.
So what about a hybrid? Why not record the data, then download it wirelessly so you don’t have to retrieve it.?
In two words -see above. The problem does not go away or get better. The wireless link now has to be remotely activated, so a transmitter and receiver is required. And an antenna. And a power supply. And a recorder. And let’s think about the data transmission characteristics.
Suppose a recorder has been running for a couple of days at a good quality. It may have accumulated 3-400MBytes of data. This now has to be downloaded. Assume the link is capable of good quality bluetooth rates, say 100kbps to make the sums easy. The time to download this data assuming uninterrupted and error-free, is 400*8*1000 kb (kilobits) /100 (kilobits per sec) = 32000 seconds or about 9 hours!
Now clearly we can hope for better than this – low power data rates are gradually climbing but even at 1Mbps this is still approximately 1 hour. Higher data rates approaching WiFi rates are available but only at significant power cost. And of course the transmitter can still be detected, and any simple electromagnetic shielding around the target will nullify the transmistter. (Chicken wire works well).
The conclusion to this is that there is no simple answer. Each situation may require a different solution and you need a toolbox of them to deliver results.
Hopefully at Somerdata we can provide you with some of these tools, advice and even the names of Companies who provide the solutions we don’t. !