-
Hi.
Has anyone seen project or a product that would visualize source of noises/sounds? Something like thermal camera but for sounds.
Few microphones, camera, software to analyze and visualize. Not sure how hard it would be. Probably not that easy on software side.
Edit: something that would allow me to visually see (like in thermal camera) which capacitor/coil is making sounds or which part of car engine is making noise. -
It's called sonar.
Presumably you're thinking about using it in air and want a decent resolution.
The problem is high frequencies are required for high resolution and are attenuated fairly quickly in air. For example the wavelength of 100kHz in air under standard conditions is 3.4mm and sound at that frequency doesn't travel very far.
Imaging with sound is more effective in water which is a better conductor but higher frequencies are required as the wavelength is longer. -
An interesting concept but much depends upon what you intend the unit to do.
Maritime Sonar running in passive mode can listen for sound and provide an indication of direction but not distance (as it is not running with an active ping that can be timed)
The detection array can be steered electronically and would likely consist of a large number of acoustic receiving elements in an array format on a flat panel. The problems is that, unlike a thermal camera imaging array, the acoustic array acts as a single pixel.
Such a system will only tell you that there is an audio (or ultrasonic) signal in view of the directional receiving array and where it is located within the field of 'view'.
It would likely be easier and more efficient to make an acoustic 'torch'. Such a unit would be equipped with a very directional mic feeding some form of audio level processing and display system. A laser pointer could provide visual targeting information. An alternative would be similar to the thermal 'torches' that actually illuminate the target in a colour representing the captured data. In the case of a thermal torch it is temperature, for audio it would be signal level. The user would scan the area of interest and the units light projector would bathe the 'target area with the appropriate colour light. The projector beam-width could be matched to the acoustic beam-width.
Theoretically you could automate the torches scan using a mechanical servo system and it would 'write' a colour display on the target area.
Aurora -
A bit like this one?
http://smins.co.kr/html/en/product/product_0201.html
It might not quite have the resolution you want and a bit expensive at £25k or so.... -
You could use a multi-capsule microphone like a SoundField or Eigenmic with some heafty DSP. You would be able to create a color graded 2D or 3D image of sounds around the mic based on level or frequency.
Though this would be really fun, it would be quite costly, though not extremely difficult since many formulae for decoding B-Format or other matrixed audio are published and well documented, ans can give you R-theta-phi coordonates for level or frequency. -
@nali,
Nice find
The unit looks very compact .....impressive. I shall have to see what can be found on its operating principles.
Aurora -
@nali,
Nice find
The unit looks very compact .....impressive. I shall have to see what can be found on its operating principles.
Aurora
I saw it on TV a few nights ago on the One Show in an article about noisy frogs, it was a "I want one of those" moments
Here's a link which contains the video clip showing it in use (about 2 min in). Being BBC not sure if it's watchable outside the UK
http://www.thp-systems.com/sm-instruments-acoustic-camera-used-to-locate-marsh-frogs-as-part-of-a-recent-edition-of-the-bbcs-one-show-television-program/
-
I've seen a video of a commercial (for military presumably) automated turret gun that fires back at the exact location a sound comes from. An array of spacial microphones does the trick.
-
I shall have to see what can be found on its operating principles.
The principles are pretty simple, although I'm sure there's a lot of hard magic involved in practice.
Take an array of sensors (microphones, antennas, whatever), and imagine them picking up a sinusoidal signal that originates directly in front of the array. The distance from the source to each sensor is the same (assuming the source is far enough away), so the signals are all in phase. Add them together and they reinforce.
If the source is not directly in front, then sensors closer to it will pick up the signal before ones that are further away. If you add the received signals together now, you'll get partial or complete cancellation. But if you know the direction that the source is in, you can compensate for the delay. Shift the phase of the signal received by each sensor by the appropriate amount, and the result is all in phase.
Since you don't know what direction the signal is coming from (and, in fact, you'll generally have multiple signals coming from multiple directions), you try them all at once. For each possible direction, apply the appropriate phase shifts, and you'll see what's coming from from that direction.
Repeat this for each frequency in the range that you care about, and you get an image.
This sort of thing is common in radio astronomy. It's also the idea (in reverse) behind that Russian radar system that was discussed here recently. -
@DJohn,
Thank you for the very clear explanation. Much appreciated.
Aurora -
@DJohn,
Thank you for the very clear explanation. Much appreciated.
Aurora
Google "Phased array" -
-
I've seen a video of a commercial (for military presumably) automated turret gun that fires back at the exact location a sound comes from. An array of spacial microphones does the trick.
They work on the principle of two sounds for each shot - the first is the sound of the shock wave as the bullet passes, the second sound is the gun shot. The two gives the direction, the elevation and the range. The one I used have 4 microphones - 3 in a triangle and the fourth was straight up.