Every sound sensor uses a microphone, because a microphone is the component that turns sound energy into an electrical signal that the sensor can measure. What determines whether a device is a privacy risk is how the rest of the system handles that signal.
In this article, we’ll explain why every sound-measuring device contains a microphone, what could turn one into a recording device, how our sensor is built so that it can't become one, and what to ask any vendor before you choose a sound sensor.
No device can measure sound without a microphone. A microphone is a transducer (a device that changes one form of energy into another) that converts sound pressure variations into an electrical signal. Every sound measurement starts with that conversion.
This is true even for a device that only outputs a decibel reading, which is a sound pressure level that measures how strongly air pressure fluctuates as sound passes through a space. Inside every microphone is a thin, flexible membrane called a diaphragm. As sound pressure pushes and pulls on it, the diaphragm moves, and the microphone turns that movement into a matching electrical signal.
Most sensors in modern devices use a MEMS (micro-electro-mechanical systems) microphone, which applies the same principle at a much smaller scale: a microscopic diaphragm built onto a silicon chip.
A loudspeaker works the same way in reverse. An electrical signal drives its diaphragm back and forth, and that movement pushes air to create sound. Because the two run on the same principle, a speaker can also act as a microphone, which is why older intercoms used a single speaker to both talk and listen.
Whether a sound sensor could become a recording device depends on how the system is built: the hardware, the firmware (the built-in software that controls a device's components), and the software around the microphone.
To assess a sound sensor’s security risk, start by asking three questions:
Ultimately, sound sensor security comes down to how each vendor builds their specific device. It's important to address these considerations upfront, so you can avoid any unwanted surprises after installation.
Many vendors will avoid saying their sensors contain a microphone because that immediately raises privacy and security questions. Those questions are well-founded, as IT and security teams need to know that the device on their wall can't be turned into a recording device.
These questions often get vague answers, given how much security engineering it takes to guarantee that audio never leaves the device, and many are not built with the necessary safeguards at every level.
At Kaiterra, we are fully transparent about how our sound sensor works: like every other sound sensor on the market, it uses a microphone, but we built in several layers of security (from hardware to the cloud) to ensure no audio will ever leave the device and your security stays intact.
Kaiterra's sound sensor has a privacy-first design, with security built into every level, from the hardware and firmware to the software and cloud, so it's physically impossible to extract an audio recording, even if every other part of the system is compromised. Here's how each layer works to protect your privacy:
The microphone switches on for 87 milliseconds once per second, then switches off, so it’s listening only 8.7% of the time - nowhere near enough to make sense of any conversation. That timing is set by physical electronic components soldered onto the circuit board, not by software, so no update or command can make the microphone listen longer or more often.
The sound sensor has its own microcontroller unit (MCU), physically separate from the rest of the monitor. Each 87-ms snippet is held in the MCU’s short-term memory just long enough to calculate sound levels, then permanently deleted. The audio is never saved or sent anywhere else for processing, so there’s no point along the way where it could be stored or intercepted. Only the measurements leave the sensor.
The sound sensor can’t be programmed remotely, or at all. Its firmware is fixed at the factory and can’t be changed by anyone, Kaiterra included. The access port that engineers would normally use to read or reprogram it is permanently disabled with a one-time hardware fuse, and the sensor’s connection to the monitor is wired to only send data out, so nothing can be sent back in.
The monitor and Kaiterra’s cloud are protected by the same security practices as every Kaiterra device, including data encryption and device certificates that verify each monitor’s identity. The full details are on our security page.
In the worst-case scenario, where both the cloud and the monitor’s own software are breached, there’s still no way into the sound sensor. Even if the sensor’s firmware itself were compromised, the hardware would still limit the microphone to 87-ms snippets, far too short and too far apart to capture intelligible speech.
The sound sensor is only capable of sending numerical readings to the monitor: six statistical sound levels every minute and a frequency spectrum every five minutes. None of these values contain audio.
|
Metric |
What it tells you |
|
LAeq |
The average sound level, weighted to match human hearing |
|
LCeq |
The average level including low-frequency energy, such as HVAC rumble, that A-weighting discounts |
|
LZeq |
The unweighted average level, used as a reference |
|
LAmax |
The loudest one-second level in the period |
|
LA10 |
The level exceeded 10% of the time, which captures intrusive sounds like speech and activity |
|
LA90 |
The level exceeded 90% of the time, which is the room's background noise floor |
|
Frequency spectrum |
Sound energy in 31 one-third-octave bands from 20 Hz to 20 kHz |
The frequency spectrum readings were also designed so they give nothing away about who is speaking or what they're saying. Adult male and female voices tend to sit in slightly different frequency ranges (roughly 85-180 Hz and 165-255 Hz, respectively), but the sensor averages the sound energy in each band over five minutes, so the data reflects the overall balance of a room rather than any individual voice.
Even without any audio, these readings give workplace and facilities teams all of the information they need to measure and maintain acoustic comfort in their spaces.
Over a single minute, a sound sensor that samples only part of the time can miss brief loud events, but over days and weeks its average level, variability, and frequency readings converge on their true values. Sampling only part of the time is necessary to ensure privacy, since the short listening window is what keeps the microphone from ever capturing intelligible speech.
As samples accumulate, that volume of data will surface patterns in how a space sounds as occupancy and activity change across the day and week, which a one-time sound check would never be able to reveal.
Before choosing a sound monitoring solution, ask the vendor exactly how the device mitigates security risks. These questions are a good place to start:
Look for specific answers that explain how each safeguard works, rather than general assurances. The strongest answers will be backed up by verified security documentation.
You shouldn’t have to compromise your team's privacy or your workplace's security to measure sound. With Kaiterra, you can securely evaluate how your office sounds and improve acoustic comfort without the risk. Reach out to our team to learn more today.