Acoustic camera — see where the noise comes from before you pay to fix it
A sound level meter tells you how loud. It does not tell you what. In a plant with dozens of valves, pumps, fans and pipe runs, or in a house whose roof cracks every morning as the sun warms it, the expensive question is not "how many decibels" but "which one" — because if that answer is wrong, you enclose the wrong machine, replace the wrong valve, or open the wrong ceiling.
An acoustic camera answers it with a picture. 128 microphones record simultaneously and a sound map is overlaid on live video, so each source appears as a spot on the actual equipment. Sources are ranked against each other, and the contribution of any one of them to the total can be measured — before and after a change. Geonoise sells neither instruments nor noise control, so the image points at the real cause rather than at whatever somebody wants to sell.
Where we are engaged
- Factories and power plants — identifying which valve, pump, compressor, fan or pipe run governs the level at the boundary or at a workstation; ranking them before money is spent, and re-measuring afterwards to prove the spend actually reduced the noise.
- Component changes and maintenance — you replaced one valve or nozzle: how much did the noise change? We measure that component's contribution to the total before and after, instead of inferring it from a boundary figure that moves for a dozen other reasons.
- Compressed air and gas leaks — leaks in compressed air, steam and gas systems appear as high-frequency sources on the image, so an entire line can be swept in hours.
- Buildings and homes — a roof that cracks as the temperature changes, an impact that travels through the structure, services noise nobody can attribute to a particular fan, or leakage paths through walls, doors and windows: located on an image before anything is opened up.
- Disputes and complaints — when the parties argue about which side the noise comes from, an image showing the source with a measured level ends the argument faster than a table of numbers, and it supports an expert report.
- Product development — your machine, enclosure, door or panel missed its rating. The image shows where the energy escapes, before the whole assembly is redesigned.
Why 128 microphones
Every acoustic camera builds its image from the differences in arrival time at each microphone. The number of microphones and the size of the array therefore decide whether two sources close together can be separated, how much quieter than the dominant source a secondary source can be and still show up, and how low in frequency the image remains usable.
- Spatial resolution — two valves a metre apart on the same line resolve as two spots, not one blur.
- Dynamic range — secondary sources well below the dominant one remain visible, which matters the moment you have fixed the dominant one and need to know what is next.
- Whole-event recording — every channel is stored, so any moment and any frequency band can be analysed afterwards. A sound that happened once at 05:30 is not lost.
- Compared with a sound intensity probe — intensity scanning to ISO 9614-2 quantifies sound power accurately, but it is swept point by point, takes hours, and cannot capture a transient. We use both: the camera to find and rank, the probe to confirm the number where a standard requires one.
The sound that happens once a day
The hardest noises are the ones that do not occur while the technician is standing there. A metal roof expanding in the first sun. A structure contracting as the evening cools. A machine that is only loud during a load change. For these we set the array up in advance and record continuously across the window in which the event occurs; the engineer then replays each event individually. The image shows whether the sound originates at a particular joint, purlin or sheet, and whether it stays at one point or runs along the frame — which is what a roofing contractor needs before touching anything, and is evidence a builder or developer finds hard to argue with.
The limits of the method — stated plainly
- Very low frequencies — image resolution falls with frequency. A low-frequency hum from a transformer or chiller is investigated with the camera alongside vibration measurement and spectral analysis, not with the camera alone.
- Sources behind a solid wall — the camera sees the surface that radiates, not the cause hidden behind it. We interpret the image together with how sound actually travels through the structure.
- Numbers for compliance — the image tells you what is where. The figure you submit to an authority is still measured with a Class 1 meter to ISO 1996-2 in the normal way, and the same report contains both.
What you receive
- Annotated acoustic images showing each source on the real equipment or structure, by the frequency bands that matter
- A ranked source table with each source's contribution to the total at the receiver or boundary
- For remediation work — before-and-after images and levels, so the decibels the measure actually delivered are on record
- A priority order for fixes by effect per baht, with no stake in what you buy
- A report signed by a named engineer, usable in an EIA submission, an expert report or a negotiation with a contractor
Standards & method
Beamforming acoustic imaging with a 128-microphone array, all channels recorded for post-analysis · ISO 9614-2 (sound intensity scanning) · ISO 3747 (in-situ sound power) · ISO 1996-2 · ISO 16283-1 and ISO 16283-3 · ISO 11820 (in-duct silencers in situ) · IEC 61672-1 Class 1 · instruments calibrated by an ISO/IEC 17025 accredited laboratory
FAQ
What is an acoustic camera and how is it different from a sound level meter?
A sound level meter has one microphone and reports the level at that point. An acoustic camera has tens to hundreds of microphones, processes the differences in the time at which sound reaches each of them, and overlays a sound map on live video — so you see where on a machine or building the noise originates, rather than only how loud it is.
Can it find a noise that only happens once a day, such as a roof cracking in the morning?
Yes — this is where the camera beats a walk-round with a probe. We set the array up in advance to cover the window in which the sound usually occurs, record all channels continuously, and analyse each event afterwards. The image shows which joint or sheet of the roof produced it, and whether it was one point or several in sequence.
Can you measure how much one valve contributes to the total?
Yes; this is the core industrial use. The acoustic image is resolved by position and frequency band, so the contribution of a selected source can be isolated. When you replace the valve or fit a silencer, we re-measure under the same conditions and report the difference in decibels, with before-and-after images.
Does it work outdoors and at a distance?
Yes. The array works outdoors and can be pointed at a plant, a pipe rack or a building façade from tens of metres away. Distance trades against resolution, so we choose the camera position to suit the size of the things that need to be told apart.
What about a low-frequency hum?
Image resolution falls with frequency. For a hum from a transformer, chiller or pump we use the camera to eliminate candidates, combined with vibration measurement on the equipment and spectral analysis at the receiver — which confirms a low-frequency source more reliably than imaging alone.
Can the new low-cost cameras do the same?
Small handheld units with a few dozen microphones are well suited to finding compressed-air leaks or a single obvious source. Their limits are separating sources close together, seeing secondary sources below the dominant one, and the low-frequency range. A 128-microphone array that records every channel gives enough data for ranking and contribution measurement, which is what an investment decision needs.
How long does a survey take?
A typical plant or building takes half a day to a day on site. A sound that only occurs at a certain time needs the camera set up to wait for it. Programme and scope come with the proposal.
Does it replace a compliance measurement?
No, and it should not. The figure submitted to an authority is measured with a Class 1 sound level meter by the prescribed method. The camera answers the next question — if the limit is exceeded, where do you fix it. The same report can contain both.
Do you have a conflict if we ask you to fix what the camera found?
No, because we do not sell or install noise control. We locate the sources, rank them, specify the remedy and re-measure to confirm it worked — but we have no financial interest in it being expensive.