In an effort to reduce the murder rate I want to greatly increase the chances of getting caught. To do so I invented a device that will record your last words (and potential the murderer's and your surroundings) that could provide key insights into your death.

This device would be a small chip inserted into the body. The chip would have a built in microphone measuring vibrations through your body produced by sound waves. The sound waves would be recorded into internal memory, small enough to hold just a ten minute loop (constantly overwritten). The device is powered by the circulatory system. When the person who has the chip implanted dies the chip is no longer provided power and thus the last 10 minutes are not overwritten. In the event of your death the chip could be pulled and listened too to give insight into your last moments, if no foul play is suspected the recording can be given to the family as a keepsake.

Is this invention technologically feasible? Can you power such a device using just the electricity produced by the body? Where in the body should it be to get the best recordings and power while being the least invasive? How big would the device be, classic CIA-eqse implant or more like a pacemaker? Finally, what kind of recordings could it get, would it only record your own voice as vibrations from your voice box would be strongest or could it pick up others talking to you? Would it sound like real voices?

  • 3
    $\begingroup$ This device would not, in any way, deter a murderer, robber, or people with nefarious intentions. Recording video through the victim's eyes wouldn't either. A simple way of hiding your identity would be to not speak while you murder your victim. You could even play a recording of someone else's voice in order to cast suspicion away from you. This whole thing is incredibly silly, and ill conceived. I mean .. at the end of the day, if I were to kill someone and really thought that the chip could hold some information to give me away, I would just cut the chip out of my victim. $\endgroup$
    – AndreiROM
    Jan 17, 2017 at 18:04
  • 1
    $\begingroup$ @AndreiROM Sure for people with well planned perfectly executed murders they would mask their voice,etc. But for random bar fights, gang violence, muggings, etc. it could still help (if it worked). That is like saying because fingerprints can be used to find a criminals identity they would never leave fingerprints again... $\endgroup$
    – DasBeasto
    Jan 17, 2017 at 18:06
  • 1
    $\begingroup$ I strongly suspect that when being murdered, my last words will be AAOOOOOOW! If the microphone is placed suitably, it may pick up various other last sounds. $\endgroup$
    – Willk
    Jan 17, 2017 at 18:15
  • 2
    $\begingroup$ @AndreiROM Well in my story it wasn't going to be so much a voluntary procedure, I would only need to incentivize the government which would be done via the lowered murder rate. I was also hoping the technology would allow for more of an "injection" type procedure rather than full on surgery. The plot indeed involves it working at first and eventually criminals becoming smarter in avoiding it as well as more intricate ways of using the recordings ("you hear that train in the background, that only passes by X location at that time"), etc. $\endgroup$
    – DasBeasto
    Jan 17, 2017 at 18:27
  • 1
    $\begingroup$ See the Neanderthal Parallax by Robert J. Sawyer for an example of a society where everything is recorded (video and audio), and the records are kept under the supervision of a supposedly incorruptible authority which authorizes limited access for use in judicial trials. $\endgroup$
    – AlexP
    Jan 17, 2017 at 23:01

1 Answer 1


The chain of devices you need is: Microphone, pre-amp, processor, memory.

The microphone is easy. The CIA has been making odd sized microphones for years.

In a paper on low power microphones, researchers at Dartmouth designed a microphone pre-amp that consumes around 50uA. They did fabricate their device using 0.5um fabrication techniques, but they did not specify the size of the chip. Regardless, it should be quite small.

The 10 minutes of storage is tricky. If we use a simple processor that just shuffles the data from the pre-amp to memory, we'll need to store all of the data raw. The pre-amp above is specced for 4kHz sound, so we need 8kHz sampling. 8 bit audio will be fine, so that's 8000byte/sec, or 4.8Mb. On the other hand, if we ran a mp3 encoder, we could cram that data into 8Kbps mp3 streams, dropping our data needs to a mere 600kB.

We're going to need non-volatile memory, because the whole point is to store the last words. That means FLASH. The SST25 series is a low-power FLASH memory chip. It turns out that the energy usage of flash is almost independent of the size of the chip, so we can use the same device for both .wav and .mp3 versions... no need to spec them out separately. This chip uses 8uA when idle, and 30mA when writing. Fortunately, we don't need to be writing at all times. We can write in bursts to minimize power usage. For the .wav version, we'll be running at approximately a 10% duty cycle, so the actual energy consumption would be 3mA. For the .mp3 version, we would be running at 1.5% duty cycle, consuming around 0.5mA. These chips run at 3V, so they consume 9mW and 1.5mW respectively.

The processor is really the make-or-break component. A ENA2351 MP3 encoder proudly announces:

This product features a built-in hardwired MP3 encoder/decoder system, enabling the industry’s lowest power consumption of 5mW...

The difference between powering a large FLASH full of .wav and powering both a FLASH full of .mp3 and a converter is pretty small. It's the difference between 6.5mW and 9mW. Because I think converting to .mp3 is a lot of trouble and the benefits are small, it makes sense to downselect here and decide to just store .wav data.

For processors, we can look at ultra-low power processors like a MSP430. This little 8MHz processor sports the Analog to Digital converter we'll need to get data from the microphone preamp, and also sips power, consuming just 0.4mA at full power (at 3V == 1.2mA), but it also supports several low power modes that may be used to decrease this.

In the end, we're looking at around 11mW of power (10.25mW rounded up to cover all of the unusual bits)

11mW is not much power, as far as the human body goes. The human body is capable of expending 20,000 times that much. However, getting 11mW parasitically is tricky. So far tools like TEGs, worn on the outside can generate 0.02mW/cm^2. Enzymatic Fuel Cells inside the body are similarly on the microwatt range for power output.

You do have access to a lot of unusual power sources, of course. If you're implanting a microphone on a person, it makes some sense to couple to the diaphragm. You may be able to get your 11mW by hooking to the diaphragm in some manner and leveraging their breathing. They're guaranteed to be breathing all the way up to their death (or shortly before their death), so it's a very reliable source of power.

Other than this incompletely specified power source, everything I've described could be fabricated on one or two chips. Everything could go on one chip, system on a chip style, or you might keep the flash memory separate because it can be mass produced better that way.

As for what it would sound like, that's open. It depends heavily on where the microphone is. If you put it in the chest, you'll find that it sounds like a doctor's stethoscope listening to your breathing. You would probably want the microphone closer to the mouth to capture the phonemes as they are created. It most likely would not catch anyone else's voice because the dynamic range would be too great. Remember, your own voice is so loud that your brain unconsciously moves the bones of your ear further apart just before speaking. That's done so the sound of your own voice doesn't make you deaf.

  • $\begingroup$ This is fantastic. All the information I was hoping for! Thanks. $\endgroup$
    – DasBeasto
    Jan 17, 2017 at 21:22
  • $\begingroup$ This is why this site is so amazing. "Not only is your idea possible, here is a preliminary design for you" $\endgroup$
    – Mike Clark
    Jan 17, 2017 at 23:20
  • $\begingroup$ We can fit a half terabyte on a MicroSD card. It doesn't seem implausible that your gadget could hold closer to 1GB, which would give you a lot more than 10 minutes worth of recording capacity. IMHO you should shoot for an hour, or ever 12 hours. $\endgroup$
    – Matthew
    Nov 21, 2019 at 16:06

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .