One of the cameras even seems to have a Fibonacci spiral of mics! Or maybe I'm just seeing the golden ratio when it isn't there (I wouldn't be the first). The sponsor is Incogni: The first 100 people to use code SCIENCE at the this link will get 60% off: incogni.com/science
Me seeing the thumbnail: Rooms do not have echoes. They have higher reverberation times than are required for the chosen usage. Reverberation is the summation of all the reflections from the walls, floor, and ceiling. The echo involves a repeating signal of a certain frequency over time which is produced by longer distances than are found in today’s rooms.
They're using color as a heatmap to indicate intensity but they could use color to map the frequencies and use saturation to indicate volume, that would look more like a camera and could capture many frequencies at the same time, avoiding the "focus" issue shown with the ukelele.
one source can generate so many frequencies, so if you use colors for frequencies, then many of them would be seen as white. When you only color the most dominant frequency, then you lose information about the other frequencies. However, I would like to see the sound that way, corresponding the sound freq to the light freq would be interesting.
@@lordofthechimie I mean yes but coherent sound typically has a couple dominant frequencies at any given time otherwise it'll all sound like white noise
I work in a chemical plant and we've actually used these to detect small gas leaks because they make a noise that we normally can't detect. It's super loud in the plant yet this device can differentiate between all those sounds and actually shows you a hotspot from were the leaking noise is coming from. It has saved it so much time looking for leaks with gas detectors :D
@@snorman1911 I've had times that the sound was so loud, if you got within a certain distance you couldn't tell where it was coming from anymore because it was everywhere. Your ears can get overwhelmed and it's super disorienting xD
Regarding audio triangulation & noise filtering, I noticed that after losing completely my hearing in my right ear I lost not only the ability to determine the direction of a sound (my hyperbolae of confusion, became a sphere of confusion :) ). But I also lost the ability to selectively concentrate on specific sounds (and filter/ignore others). I didn't realise I was doing it normally (before losing hearing in one ear) as it happens unconsciously. But apparently the ability to at least roughly determine the direction of a sound allows the brain to differentiate between them easier and thus to focus more on one sound while (partially) ignoring the rest. Now that all sounds come together in a single ear, I can't do either, which in a noisy place can be a problem. Many people probably know (though not so many have realised) that concentrating on specific sound (like simply looking at whomever is speaking to you) in a noisy environment allows you to pick details (like articulation) in that sound better (and of course we apparently can "read lips" unconsciously to a tiny degree, though that's really tiny factor. Just enough sometimes to the brain to pick a sound among several similar, because of the shape of the mouth of the speaker).
I have also lost hearing almost completely (maybe 90% or so) in my left ear, conversing in public places or trying to pay attention during conferences is not fun :( Many times I can even hear the words but the overlapping sounds make me want to repeat it in my head to fully process them.
@@SahilVerma-wm6ie , I'm sorry to hear that! Yes it definitely isn't fun :/ The noisier the environment is the more effort I also need in order to (re-)construct words & sentences. But louder noises at key words sometimes can devoid whole sentences of meaning. Basically, above certain noise threshold it's not really worth the effort. I simply make sign (or shout back) that it's too noisy and just ignore most verbal communication (and consequently such places in general).
This is obvious in that people can determine direction with a single ear. Same with the stupid claim that you need two eyes for depth perception. NO YOU DON'T. A person born with one working eye still has depth perception. A person that covers one eye doesn't lose depth perception. It is honestly quite sad that so-called experts make such blind (deaf) assertions, especially given the fact that AI, machine learning, and neural networks exist on computers now. The brain makes an association based on data received, adjusts that data with what is expected, and gives an output that is fed back into the system. Thus, with one ear, if a person hears a sound associate with event A, it can determine direction via associating other things, and the 'neural network' refines itself as such. This is INCREDIBLY OBVIOUS, yet the naïve and stooooopid explanation that "the brain uses the distance between your eyes to calculate depth; the brain uses time delay of ears receiving sound to calculate direction." NO IT DOESN'T. The brain isn't a calculator (at least, not in this regard). The brain isn't doing this calculation at all. I can determine direction with a single ear; it isn't hard, so I am at awe that you supposedly cannot do this.
I’m curious as to how that could be improved as I could see there being “tricks” or Methodes that would allow you to do this better. My grandpa (early 80s) has the same issue, simply because he hears little in both ears. So your problem I believe is common in people with hearing problems. Best of luck m8!
I have two working ears but my brain finds it difficult to filter noises so I have the same problem in crowded places. It definitely helps to be able to see the person's mouth. I struggled a bit when everyone had to wear masks!
The company I work for (rolling stock maintenance) recently bought the Fluke industrial imager 900. It is a bit pricey, but absolutely invaluable when searching for airleaks on full trainsets. A job that one year ago could take a couple of days for a couple of people is now down to a one-man job in 90 minutes. Not only is it time-saving, we're also finding air leaks that otherwise would have been unlikely to find at all, due to hidden placement (hard to reach with soapy water) and frequencies not audible by human ears. It's like a miracle tool that we've been dreaming of for decades.
@@SteveMould Hi Steven! Have you thought about the idea of laser microphones? The CIA (supposedly) developed a few, but can't find much on the internet It was in a Splinter Cell game too: two bad guys were talking in an elevator, and the theory is that a flat surface resonates from their voice, and the small resonations can be detected from afar by a laser beam
I remember hearing years ago that a research group designed a way to put hundreds of microphones around the perimeter of a sports arena and then fed all those signals into a computer. This allowed them to select a single person out of a stadium full of people talking and eavesdrop in on their conversation with perfect clarity even if they were talking quietly. Amazing to think what they can do now with so much faster tech.
We visualizes the sound of 75000 football fans in FC Bayern Munich’s Allianz Arena. Look for the project "REIMAGINE THE GAME" - it's a pretty revolutionary football experience.
There are so many things you could do with this technology, but as someone who dabbles in music, the first thought I had was recording a band playing in a room, and replaying it through vr like a virtual concert, but you can walk up to each person and hear their instruments get louder and quieter as you move towards and away from them
I think the problem with that is processing speed on the user’s computer. To make it sound good you need reverb on all the instruments, and that reverb needs to match up to where you are in the room. Modern computers can’t handle that many reverbs unless they’re really short
I like your concept a lot. So ignore the comment below saying you can't and see if you can turn your idea into something. Some computer games already do this, and as you move around a 3D world, where sounds appear to come from changes in direction and loudness.
There are much easier ways to make this happen Just record each instrument to its own track and play it as a sound source in the room.... you know, like literally every VR game I think a teenager could whip it up in a weekend, and an experienced dev could do it in an hour
3:36 *_SOUND WARNING FOR PETS!!!_* My cats both freaked out hearing that sound! One of them bolted for cover and the other one shot up and stared at me like I fired a gun off hahaha. My phone wasn't even at full volume, and it's speakers are kind of weak too. So... watch out for that, especially if you have a cat _in your lap_ at that point in the video (RIP my thighs).
It is so cool to see the order of reflections from the clap in the room…having built studio control rooms and imagining what the reflections look like. Before complex acoustical modeling was available, mirrors on a first reflection wall could be used to see the sound source (monitor speaker) to determine where to install sound absorption. The sound camera is so practical for this and many more applications.
I scroll down the comments to look for a music recording production related one like this one. My thoughts about this technology was in terms of using it to design and build the perfect mixing room. Where bounces of walls, materials, space dimensions and reflections are perfectly aligned, final result in a true to real transparency in the mix.
This is exactly what my first thought was with the echo. Then I looked up the price of the cheapest acoustic camera and concluded it would be cheaper to just install acoustic dampening over the entire walls.
Banger video, great job Mr. Mould! I recently took an acoustic signal processing course (I'm not an audio engineer though), and I believe the functional principle here is called "beamforming". It's a conceptually straightforward principle, and most phones and computers today have multiple microphones to make use of it. That's how some laptops know to wake up when you sit in front of them, but stay sleeping when random noises are made in the same room.
2:16 Ok, this is legitimately amazing, how come I've never known about this? an acoustic camera? that's just plain cool, and I'm kind of impressed that its possible at all.
And yet the maths, although computationally complex, is actually quite reasonable. It's a really clever application of science and really useful. It's engineering solving problems.
You should've mentioned that a 2-mic or 3-mic version of this is how noise cancelling microphones work (like on your phone during a call). You use the delay between the mics to localize the direction every sound is coming from. Increase the gain from sounds coming from one direction (where the person's mouth should be if they're holding the phone against their face), decrease the gain from sounds coming from all other directions.
A classmate and I built a very basic acoustic camera back in 1972 as a final year project in Physics at Aberdeen University. We took still images in a darkened room of a single red LED changing brightness as it moved through a sound field. Not bad given that red LEDs were the only colour available at the time and had only reached the market in 1968
@@brucemcpherson8832 ok. But UA-cam is social media. Even more if you interact with the comments. Usually people that wasn't raised in this trans digital environment don't bother (and they're in their right to do so) to invest on this behavior. I may sound like an alien because I'm not a native English speaker and also I tried to sound as neutral as possible lol
oh! I've had the idea of doing this with Wifi signal strength - you could write an app that changes the screen colour based on the signal strength, then do a long exposure shot while moving the phone around. It would require some patience, however!
I think it does not make a difference how high you are on the ground, the other point of intersection is very very high, about twice the height of the satellite, no?
I always love stuff involving audio and sound, resonant frequencies, and the fact that we can "extract" frequencies / filter specific frequencies out and stuff is wild. I feel like audio based electronic sensors and such have such a massive future. being able to map / pinpoint the location off sound and how that could progress to audio in 3D space is wild.
This would work great in orchestral or unamplified acoustic scenarios, but in live amplified performances, which is most concerts, it's better to just use the raw audio from the instruments and then mix and pan that with a recording of the crowd to create a 3D soundscape
The way these acoustic cameras work really reminds me of how interferometry radiotelescopes work, i would bet its the same principle (a large array of omnidirectional detectors working together to create a 3d image)
it is exactly the same thing, it's just an acoustic analogy, a lot of the same principles are exchangeable. Another very cool thing that works in the same way is beamforming with loudspeaker arrays but instead of listening in a direction you can project sound to a very narrow direction. this is what most sound bars do and what the leviathan v2 pro does using black magic
We use these cameras to find oxygen leaks on airplanes. When isolating the correct frequency, we can see leaks we wouldn't have found without hours of using bubble solution to many lines and connections or waving special microphones around inside the plane which only work close to the leaks. When pressurizing planes, these also show the locations of bad seal leaks, and inside electrical panels I've found sources of arcing or they help isolate individual chattering relays within rows of them. For airplane maintenance, they are fantastic!
While the industrial and appliance applications/overall equipment monitoring capabilities certainly seem useful, when I saw that "eye dropper" tool being used to isolate certain sound sources, I couldn't help but consider the implications for audio production/sound design applications. Would be a dream to have something like this in a recording studio.
My thoughts exactly but applied to programs broadcast on television - isolate certain sound sources - my bugbear is overly emphasised music track competing with dialogue or voice over - I would love to kill the music and just listen to voice track. And then there are commercials - industry produces cue pips on programmes for breakaway commercial playout which are not incorporated into the tx signal - if it were you could mute top and tail of commercial breaks instead I indulge in timeshift on my PVR.
Why do you film in 60fps in the UK? Shouldn't you want to film in PAL (50fps) to interact better with the electric grid if/when you have any lighting in shot so it doesn't produce a banding effect? I had so many issues with this when I first moved here, but now I shoot all my videos in 50fps (or 25 for my second channel) and it's grand :)
Yes! But because most displays that my videos will appear on will have a refresh rate that is a multiple of 60, I’ve chosen to film 60 to avoid that judder. I’ve set up all my studio lighting in a way that avoids the 50hz mains issue (my led panels don’t flicker). When I’m filming indoors away from my studio a switch to 50
As a Sound Engineer for music recording studios this must be a huge success (or will be). Making a room "flat" by using material to absorb and/or diffuse sound waves, hence reducing "nodes" - points where reflected sounds of the same frequency collide and become louder and hence create an unbalanced listening space - should now be very easy. Anyone in the biz experienced this yet?
I actually see patterns like the echoes at 9:50 when I’m listening to clapping in rooms with off-white painted cinder block walls, it looks like black and white visual noise as a faint overlay. I always thought I was imagining it, especially since I can only resolve it on certain backgrounds and in rooms made of certain materials (and the room needs to be ~5 to 10m per side). Looking at these though, it seems like, crazy or not, the stuff I’ve been seeing may have some connection to reality.
You’re most likely seeing infrasound! Our eyes resonate just below our hearing range, about 18-19Hz. The wavelength of such a sound in air is about twice the room size you mentioned, so I figure that when you see this, your eyeballs happen to be near an antinode of a standing wave caused by the echoes from opposite walls interfering. And an offwhite cinderblock wall just happens to reflect a lot of both sound and light, so you can better see the visual disturbance caused by the sound deforming your eye. I don’t know why that particular frequency-I assume it has something to do with the acoustic impedance of the vitreous humour-but still, it’s a well documented phenomenon. I first learned of it in Mary Roach’s book “Spook” where there was a case of a faulty fan in a storage room causing people to see these “ghost” illusions.
@@kotresh That’s true-and in fact I have synaesthesia-but as a neurological phenomenon, it’s more like “hearing G♯m on a piano _feels like_ seeing something orange” for instance; it doesn’t affect the actual sense. Whereas this has a perfectly good explanation as a physical phenomenon, which anyone could experience, it’s just that it takes very specific conditions to be noticeable.
If you'd like to check out a different camera type, take a look at Cavitar and their Cavilux product line. (There may be other products in this space, I've just never seen them). Basically they use a very high intensity illuminator to bring the ambient light level up to things like welding arcs and explosives going off so that a camera can film everything without requiring huge dynamic range. Pretty interesting!
Oh wow, that‘s a nice surprise. I’ve been following your channel for years now, and totally did not expect to see stuff that I‘m (somewhat) working on. Great video!
Wow. Never heard of an acoustic camera before, and now that I have.. Of course there's such a thing! It's both a blindingly obvious idea and brilliant in execution. I remember a Tom Scott video about a Swiss shooting range, where they used microphones to determine how accurate each shot was - basically a simpler version of the same thing - and for some reason I never followed that train of thought to this end result. I guess that's why they make videos and I just comment on them.
Related to shooting as well, there's a microphone array device called a "Boomerang" that can be mounted to military vehicles and is capable of detecting the direction of incoming gunfire.
Fun fact hyperboloids are also used by TDOA (time difference of arrival) RTLS (Real Time Location Systems) that use UWB (Ultra-Wide Band) RF for location of a beacon. While AOA (angle of arrival) systems use the phase offset at each antenna in an array to determine angle. Definitely very cool to see what they are doing with microphones!
7:53 This visual gave me the idea of creating a program that tracks a point or a group of points to the ears and connects to another program converting those values into a template for audio. Allowing artists to create "8d" audio much easier, this could possibly change how we experience music and how it is developed, like when the drum machine was created.
There's actually a reason I didn't! What I was saying while playing the ukelele was about how you can use the camera to isolate just the sound you want (like cropping an image). But we never got that to actually work sadly
I was weaking my headset. My wife, in another room, said, "does anyone else hear that awful sound?" I just said, "what sound?" And she said, "never mind, I don't hear it anymore."
I saw a demonstation of a similar device in an industrial fair like 5 years ago, being an engineer I was thinking to a miriad of possible applications, they told me the bulk of the work they were doing at that time was basically "hole plugger" for like high end cars... being unusually noisy... it helped find the rubber plug someone, for some reason, had forgotten to install in a much shorter time than checking the whole car
Hey Steve, great video. This touches upon a product idea that I don't think exists, but I really think it should. For people who ride with a helmet, being able to hear your surroundings is important when riding in traffic, but the sound of the wind will drown out most noises starting at 15mph. I want something that is part noise cancelling headphones, and part hearing aid. I want an array of microphones to create a noise cancelling experience and then filter all of the wind noise out and only allow environmental noise through. To my surprise this still doesn't appear to be a product. I'm under the impression that all the technology to make this product exists, but this product doesn't exist, so maybe I am missing something? i.e. I want headphones/helmet that can give me super hearing.
Hearing protection made for shooting guns have this feature. They amplify ambient sound while suppressing pressure spikes in the audio. Howard-Leight makes a cheap pair that are popular with firearms enthusiasts
a traditional software based sound processor (named DSP, or Digital Signal Processing) subtracts white noise on mic input, but it also subtracts traffic volume, why? because the traffic and white noise overlap in frequency so both is subtracted, so the traffic volume is a fraction of original and has to be amplified to be audible, but the white noise is random, so some noise escape the subtraction and is amplified, creating weird sound. The keyword is "traditional ways", maybe future noise cancelling uses AI for reconstruct traffic sound while leaving out white noise.
@@xponen You sound like you have expertise in this area and I value your input. I want to be clear what I mean by traffic sound. If it is a directionless traffic drone noise, subtracting that is fine. I just want to be able to hear signals that have a clear directionality to them, which is where the array of microphones comes in. The nearest microphone will pick up a noise first, but the system filters it by default until the farthest microphone picks up the noise and the system is able to determine that it is the same signal coming from a distinct direction.
Tried once Flukes hand-held acoustic camera at technology expo. That was very impressive considering, the whole thing was just a bit larger than regular DSLR
This reminds me a bit of when we were installing the subwoofer my husband built. He carried it around to various spots, then we walked around as he manipulated the angle of the speaker cone. We were able to pretty much eliminate null spots that way. We live in a log home, so the acoustics are very, very lively. All the bare wood reflects sound waves quite well.
4:30 I don't think you can actually compare the sample rate to the frame rate, right? The camera is unique in that for each ""pixel"" (lots of quotation marks) you get a full spectrum, but it would just be like a hyper-spectral camera (cameras where each pixel stores the information of the full spectrum, which for visible light it's even in THz, nothing compared to sound KHz in that sense). I'd love if you could make a video about them btw! :D The frame rate is something quite unrelated to the sample rate
Correct. The Fourier transform shows how the location over time and the frequency are related. The range of frequencies you can capture depends on the sampling rate (Nyquist-Shannon sampling theorem) and the resolution of those frequencies depends on the number of samples you take. Two samples back-to-back tells you how the air pressure is changing at each microphone, but it doesn't give you a frequency, since that's a property of a waveform over time. Only with many samples can you separate out those variations into frequencies.
I remember in 3rd grade, my best friend's dad told me about an idea for an invention. Glasses that you can put on your face and see sound as a visual. He said it would have all kinds of colours for different frequencies. It's crazy to think that by the end of my lifetime, I might actually see that entire technology develop to his seemingly impossible idea from 2005.
If anyone finds it difficult to see why you need 4 mics to locate the sound, think of it this way: Imagine 2 intersecting spheres, the intersection they make is a ring. Now push that ring into a 3rd sphere, and the ring will contact the 3rd sphere's surface at 2 points still, so you need a 4th sphere.
Or you can think of it like triangulation I understand the cones and spheres are more accurate to how it actually works, but understanding it as distances between points describes it identically, without the waffle.
Just a note, at 0:40 when you're talking about Automatic camera exposure, you've set it to Aperture mode, not Automatic. Being unfamiliar with that particular camera, I THINK if it has an automatic mode, it would be two more clicks, just after P (for Program) the little A inside of camera symbol. On a digital camera, it could be argued that automatic exposure would be relevant on S (shutter) A (aperture) or P (program) since the camera automatically adjusts settings on those modes to get exposure correct. And automatic ISO could even apply to M (manual) mode as well! But the 'A for Automatic' that's being insinuated isn't entirely correct :]
This reminds me of when I was a bicycle mechanic and would listen to customers rolling their bikes into the shop and use the sound in my diagnosis. I can still diagnose passing bikes which sometimes gets annoying.
in 4:02 , you can see steve suffering the horrible high-pitched torture. and still see signs of pain when the camera goes acoustic mode.... for your dedication to science i solute you...
Interesting use of the acoustic camera to detect faulty wheels on trains. Due to the unpleasantness in Ohio, we now know that thermal cameras are often used for the same purpose (though they more often detect brake faults). Related note: if your train wheel is on fire, it might be time to service the brakes.
The colloquial name for those detectors is "hotbox detector." That name dates from the days when axle bearings were something like a Babbitt bearing encased in a "box" packed with oily rags. When the bearings wore out, they could very easily catch fire and, as many freight cars still had lots of wood in their construction, that could be quite catastrophic. These days, roller bearings are required for cars used in interchange, so they wear out less frequently and there's not quite so much flammable material nearby when they do, but as we've seen, it still happens. And, as you say, they also detect stuck brakes. They're mostly not thermal cameras, though - the size and construction of a wheel doesn't change much, and the train itself provides the necessary scanning motion, so a simple passive IR sensor pointed at the spot where an axle goes by is sufficient. Because they can be so simple, they've actually been around for a long time - Wikipedia says since the 1940s.
@@RonParker interesting. Do you know if there were thermal cameras in use on the Norfolk train? I could have sworn I saw the actual footage - or at least the temperature readings - from the train. One reading shows it elevated, but not enough to trigger an alarm, and by the next reading the thing was already energetically on fire.
I went to an art installation in Vienna once that had fixed piezo sensors in a room to achieve the same effect. They had video projectors projecting lines onto the edges to features in the room, and as you made a noise they would ripple the walls etc. in that section of the room. Very eerie!
1:52 i hate that you have to hold the button when using the dyson. idk if they know but toggle buttons are a thing and they are easier for my index finger muscle
@Steve Mould about what you said about the vibrating aluminium bar/rod. Although it's possible for the vibration to be longitudinal the fact that the sound originates at the bar/rod ends does NOT imply that the vibrations are axial. Since the rod is constrained at its midpoint (acting like a node) and not at the 2 ends it can not behave like a guitar string(what you show is the 1st mode for a bar pinned at both ends), but can still vibrate in flection with the first bending mode having the largest displacement at the ends (like each prong of a diapason). That said the bar was behaving one way nor the other? A reasonable assumption is usually that the mode with the lowest natural frequency is the main one(and usually bending modes are always lower than axial because usually axial rigidity is higher) but given the way you excite the bar ("pinching" on both sides) it's possible that the bending mode is not excited at all, while the axial is. Ho To tell the 2 possibilities apart? First thing that comes to mind for someone who was taught how to without this amazing equipment would be "Frequency" : figure out with formulas the frequency for each possibility and confront with the measured frequencies. But since we have this "sound video" the source for bending mode would be more distributed along the bar with the max at the ends gradually getting less and intense towards the constraint. Axial instead originates only at the 2 cylinder bases(apparently what we see in the clip). Also Intuitively: since I mentioned diapason...this bar is aluminium and despite having a larger diameter than the usual diapason prong I assume it's far less rigid in bending than a steel diapason prong, and the length is likely 3or4 times one of a standard diapason (that is already quite high pitched) longer length and lower rigidity bring the natural frequency down so there is no way this bar would produce so much higher pitch sound than a standard diapason
Whenever someone asks me "what kind of super power would you like to have?" I'd always say that I would love the ability to see sound. Like think about everything around us that makes sound, from our fingers tapping onto a keyboard or even the sound of clothes rubbing together as we walk. Just so much stuff producing sound all around us!
We use the FLIR acoustic camera to identify leaks in compressed air systems. What is great is not only does it locate the source of the leak, but software can also identify the volume of the leak. So after mapping an air network we can say for example "you are spending $50k per year in air leaks, they need to be fixed". Compressed air is very expensive to make at an industrial scale and air leaks are a big issue in large buildings.
Timing is one thing, but the change in db is also an effect which can help us decide if the sound is near or far (if the change in db is big or small between our ears). There is probably more information available to our ears as well, since our ears are not uniform, meaning a sound actually going "behind" has a different curve then a sound moving "infront" of the ear or if it is.
@Steve Mould, the ear isn't a simple analogue-to-digital converter, so the brain doesn't receive waveforms to analyse. The cochlea performs spectral analysis, the output of which is conveyed to the brain.
Hi, great video! I’m lucky to work with acoustic cameras on a daily basis. I live in Poland where people are price sensitive but thankfully there are acoustic cameras like CAE SoundCam and others which are affordable, have high quality and performance. I can enjoy working with them everyday. It's the really helpful technology for engineers!
I worked on these issues with analog technology more than 30 years ago , the renderings were vector graphic monochrome however my focus was on AI sensory integration for environment object characterization .
I saw this idea on a website that collected weird ideas many many years ago, & I’ve always wondered why this great invention hadn’t been created ‘yet’ ? The application that I was thinking, would be absolutely fabulous, for this would be to put on a balloon and float it over a forested area and listen for big foots .
4:30 The time-domain sample rate can be 48, 96, 192kHz or what have you, but the live processing is almost certainly not being done at the same rate. They're typically done on "blocks" or "frames" of samples which are buffered up. If you wanted to process a full frame for every sample coming in you could. You would end up with a ton of frame overlap and it would be crazy inefficient and probably unnecessary.
As a synaesthete tortured by constant noise, i need this acoustic camera. 6:56, finally! This is exactly what I need, a device that can visually indicate the location of a sound source.
5:39 hyperboloid of confusion does not cause a problem for careful brains because the moment you enter an environment, your brain creates an audio-3d map of the enivronment using how which area of the room changes the sound. let me clarify; for ex: rear side of the room amplifies lower frequencies and upper area has less echo or front of the room amplifies the high frequencies etc. since your brain picks this up instantly, when you move the sound around, you can locate it not only by amplitude but also using frequency analysis. this is also how proper surround mixing to stereo is done, (only by experts) forget only playing aournd with amplitudes or phase differences. the whole tone should show a difference also not only in frequency but in time base as well, as in some frequencies takes longer to diminish in an environment due to the that environments natural resonances.
6:58 Trinnov AV receiver uses the similar technology to identify the speaker location in 3d space and optimise the sound in real time while other receivers simply can't . I didn't know these techniques can be used for other purposes than a theatre. Amazing video👍🏾
Acoustic cameras are really cool! We got a couple of setups at my university. The analysis methods using crosscorrelation are already pretty interesting, but there are methods to get much more resolution out of the array. You can use inverse convolution with the point spread function of the array to reduce the interferences at high frequencies and also get better results for lower frequencies. They were also used to compare the noise made by flying owls and other birds, to see which mechanism is responsible for these noises and so on. that is really state of the art technology right there.
the metal rod sound is the first time I've ever turned down my volume due to sheer panic. at first when you said headphone warning I was like: "nah it's probalby not that bad" but then when you did it I jumped in my chair and instantly turn down the volume all the way XD
this is really cool! I bet there's a lot of complex logic to interpret these signals. Because the speed of sound is, well, finite, and the camera is super slowmo, when the camera registers something it has already happened a while ago. This is specially relevant when looking at those echo blotches.
What's amazing is that our human ears are able to pick out direction, elevation, frequency, volume and motion (doppler shift)... with only 2 ears. What's really amazing is how much better Owls, Whales, etc, are with their sense of hearing. We need a battery of microphones and computer software to try and come close to replicating what they can do naturally.
One of the cameras even seems to have a Fibonacci spiral of mics! Or maybe I'm just seeing the golden ratio when it isn't there (I wouldn't be the first).
The sponsor is Incogni: The first 100 people to use code SCIENCE at the this link will get 60% off: incogni.com/science
.
Maybe you're on the verge of some epiphany like the guy from that movie Pi.
Adding irrational numbers is a great way to make sure your microphone spacing is easy to model, and non-repeating
Now replace all the microphones with antennas and use the EM field to view objects.
Me seeing the thumbnail: Rooms do not have echoes. They have higher reverberation times than are required for the chosen usage. Reverberation is the summation of all the reflections from the walls, floor, and ceiling. The echo involves a repeating signal of a certain frequency over time which is produced by longer distances than are found in today’s rooms.
They're using color as a heatmap to indicate intensity but they could use color to map the frequencies and use saturation to indicate volume, that would look more like a camera and could capture many frequencies at the same time, avoiding the "focus" issue shown with the ukelele.
That's what I imagine would be more intuitive. It'd seem more like it models the sound.
one source can generate so many frequencies, so if you use colors for frequencies, then many of them would be seen as white. When you only color the most dominant frequency, then you lose information about the other frequencies. However, I would like to see the sound that way, corresponding the sound freq to the light freq would be interesting.
@@lordofthechimie I mean yes but coherent sound typically has a couple dominant frequencies at any given time otherwise it'll all sound like white noise
I think simply having different color map settings for different use cases would be best, honestly.
I think the problem with comparison to a light camera, is the number of pixels. 50 cameras can't replicate millions of pixels
I work in a chemical plant and we've actually used these to detect small gas leaks because they make a noise that we normally can't detect. It's super loud in the plant yet this device can differentiate between all those sounds and actually shows you a hotspot from were the leaking noise is coming from. It has saved it so much time looking for leaks with gas detectors :D
Hi, I work in the same industry and would like to know more. Can you direct me ?
@@stingrayphelps848 look for "FLIR ultrasonic leak detector"
@@stingrayphelps848 Fluke ii900
That's awesome! The number of times I've wandered back and forth chasing the source of a high frequency sound...
@@snorman1911 I've had times that the sound was so loud, if you got within a certain distance you couldn't tell where it was coming from anymore because it was everywhere. Your ears can get overwhelmed and it's super disorienting xD
Regarding audio triangulation & noise filtering, I noticed that after losing completely my hearing in my right ear I lost not only the ability to determine the direction of a sound (my hyperbolae of confusion, became a sphere of confusion :) ).
But I also lost the ability to selectively concentrate on specific sounds (and filter/ignore others).
I didn't realise I was doing it normally (before losing hearing in one ear) as it happens unconsciously.
But apparently the ability to at least roughly determine the direction of a sound allows the brain to differentiate between them easier and thus to focus more on one sound while (partially) ignoring the rest.
Now that all sounds come together in a single ear, I can't do either, which in a noisy place can be a problem.
Many people probably know (though not so many have realised) that concentrating on specific sound (like simply looking at whomever is speaking to you) in a noisy environment allows you to pick details (like articulation) in that sound better (and of course we apparently can "read lips" unconsciously to a tiny degree, though that's really tiny factor. Just enough sometimes to the brain to pick a sound among several similar, because of the shape of the mouth of the speaker).
I have also lost hearing almost completely (maybe 90% or so) in my left ear, conversing in public places or trying to pay attention during conferences is not fun :( Many times I can even hear the words but the overlapping sounds make me want to repeat it in my head to fully process them.
@@SahilVerma-wm6ie , I'm sorry to hear that!
Yes it definitely isn't fun :/
The noisier the environment is the more effort I also need in order to (re-)construct words & sentences. But louder noises at key words sometimes can devoid whole sentences of meaning.
Basically, above certain noise threshold it's not really worth the effort. I simply make sign (or shout back) that it's too noisy and just ignore most verbal communication (and consequently such places in general).
This is obvious in that people can determine direction with a single ear. Same with the stupid claim that you need two eyes for depth perception. NO YOU DON'T. A person born with one working eye still has depth perception. A person that covers one eye doesn't lose depth perception.
It is honestly quite sad that so-called experts make such blind (deaf) assertions, especially given the fact that AI, machine learning, and neural networks exist on computers now.
The brain makes an association based on data received, adjusts that data with what is expected, and gives an output that is fed back into the system. Thus, with one ear, if a person hears a sound associate with event A, it can determine direction via associating other things, and the 'neural network' refines itself as such.
This is INCREDIBLY OBVIOUS, yet the naïve and stooooopid explanation that "the brain uses the distance between your eyes to calculate depth; the brain uses time delay of ears receiving sound to calculate direction." NO IT DOESN'T. The brain isn't a calculator (at least, not in this regard).
The brain isn't doing this calculation at all.
I can determine direction with a single ear; it isn't hard, so I am at awe that you supposedly cannot do this.
I’m curious as to how that could be improved as I could see there being “tricks” or Methodes that would allow you to do this better. My grandpa (early 80s) has the same issue, simply because he hears little in both ears. So your problem I believe is common in people with hearing problems. Best of luck m8!
I have two working ears but my brain finds it difficult to filter noises so I have the same problem in crowded places. It definitely helps to be able to see the person's mouth. I struggled a bit when everyone had to wear masks!
Imagine you fart and your camera snitches on you
There are thermal cameras which have seen this actually
The company I work for (rolling stock maintenance) recently bought the Fluke industrial imager 900. It is a bit pricey, but absolutely invaluable when searching for airleaks on full trainsets. A job that one year ago could take a couple of days for a couple of people is now down to a one-man job in 90 minutes. Not only is it time-saving, we're also finding air leaks that otherwise would have been unlikely to find at all, due to hidden placement (hard to reach with soapy water) and frequencies not audible by human ears. It's like a miracle tool that we've been dreaming of for decades.
This is somewhat reminiscent of the video cameras that "amplified motion" to stare at industrial equipment vibrating.
You could use those to find out where different sounds were coming from as well
Yeah eulerian magnification! Also can use cameras to pickup sounds from vibrating objects, Steve Mould (or Veritasium) made a video on that as well!
there's no reason at all why you couldn't apply the same processing techniques to the video from an acoustic camera setup and have both
Pretty sure steve made a video on exactly these devices
@@samrusoff he sure did
I made an acoustic camera as a dissertation project 2 years ago, really interesting to see you do a video in it!
Oh nice!
I don’t believe the university ever published it, but I can point you to some similar projects that I found during my research.
@@SamBebbington I'd love to see this if you're happy to share please
@@SteveMould Hi Steven!
Have you thought about the idea of laser microphones?
The CIA (supposedly) developed a few, but can't find much on the internet
It was in a Splinter Cell game too: two bad guys were talking in an elevator, and the theory is that a flat surface resonates from their voice, and the small resonations can be detected from afar by a laser beam
how much did it cost to make roughly
I remember hearing years ago that a research group designed a way to put hundreds of microphones around the perimeter of a sports arena and then fed all those signals into a computer. This allowed them to select a single person out of a stadium full of people talking and eavesdrop in on their conversation with perfect clarity even if they were talking quietly. Amazing to think what they can do now with so much faster tech.
We need this in both chambers of congress and we need the audio accessible by anyone over live internet streaming.
@@stefanl5183 This is good once.
NSA: write that down, write that down! 🧥✍️
We visualizes the sound of 75000 football fans in FC Bayern Munich’s Allianz Arena. Look for the project "REIMAGINE THE GAME" - it's a pretty revolutionary football experience.
I’m sure this will never be used for any spying or surveillance
There are so many things you could do with this technology, but as someone who dabbles in music, the first thought I had was recording a band playing in a room, and replaying it through vr like a virtual concert, but you can walk up to each person and hear their instruments get louder and quieter as you move towards and away from them
I think the problem with that is processing speed on the user’s computer. To make it sound good you need reverb on all the instruments, and that reverb needs to match up to where you are in the room. Modern computers can’t handle that many reverbs unless they’re really short
I like your concept a lot. So ignore the comment below saying you can't and see if you can turn your idea into something. Some computer games already do this, and as you move around a 3D world, where sounds appear to come from changes in direction and loudness.
it could be done by just recording the instruments individually, there's no need for a camera like this
There are much easier ways to make this happen
Just record each instrument to its own track and play it as a sound source in the room.... you know, like literally every VR game
I think a teenager could whip it up in a weekend, and an experienced dev could do it in an hour
you can do this without the use of audio cameras, as long as each instrument has its own dedicated mic.
You, The Action Lab and Veritasium are my favorite science channels
3:36 *_SOUND WARNING FOR PETS!!!_*
My cats both freaked out hearing that sound! One of them bolted for cover and the other one shot up and stared at me like I fired a gun off hahaha. My phone wasn't even at full volume, and it's speakers are kind of weak too. So... watch out for that, especially if you have a cat _in your lap_ at that point in the video (RIP my thighs).
Thanks bud, I was watching this video in bed with my cats
Hah, sound warning for humans too.
All 4 of my cats simultaneously woke up and looked at me with disgust from across the room then proceeded to lick themselves in unison...
And even wearing headphones, that's gonna hurt
Yeah lol mine copped it
It is so cool to see the order of reflections from the clap in the room…having built studio control rooms and imagining what the reflections look like. Before complex acoustical modeling was available, mirrors on a first reflection wall could be used to see the sound source (monitor speaker) to determine where to install sound absorption. The sound camera is so practical for this and many more applications.
I scroll down the comments to look for a music recording production related one like this one. My thoughts about this technology was in terms of using it to design and build the perfect mixing room. Where bounces of walls, materials, space dimensions and reflections are perfectly aligned, final result in a true to real transparency in the mix.
This is exactly what my first thought was with the echo. Then I looked up the price of the cheapest acoustic camera and concluded it would be cheaper to just install acoustic dampening over the entire walls.
Banger video, great job Mr. Mould! I recently took an acoustic signal processing course (I'm not an audio engineer though), and I believe the functional principle here is called "beamforming". It's a conceptually straightforward principle, and most phones and computers today have multiple microphones to make use of it. That's how some laptops know to wake up when you sit in front of them, but stay sleeping when random noises are made in the same room.
2:16 Ok, this is legitimately amazing, how come I've never known about this? an acoustic camera? that's just plain cool, and I'm kind of impressed that its possible at all.
You already have, if you watched the Dark Knight.
And yet the maths, although computationally complex, is actually quite reasonable. It's a really clever application of science and really useful. It's engineering solving problems.
@@richardcorfield9926 that's what its meant to do!
You should've mentioned that a 2-mic or 3-mic version of this is how noise cancelling microphones work (like on your phone during a call). You use the delay between the mics to localize the direction every sound is coming from. Increase the gain from sounds coming from one direction (where the person's mouth should be if they're holding the phone against their face), decrease the gain from sounds coming from all other directions.
A classmate and I built a very basic acoustic camera back in 1972 as a final year project in Physics at Aberdeen University.
We took still images in a darkened room of a single red LED changing brightness as it moved through a sound field. Not bad given that red LEDs were the only colour available at the time and had only reached the market in 1968
I love it! That's an excellent idea!
Are you 65+ years old? Kudos to you for being active in social media
@@okarowarrior I'm 72 years old, but Twitter, Facebook and those sorts of social media do not interest me
@@brucemcpherson8832 ok. But UA-cam is social media. Even more if you interact with the comments.
Usually people that wasn't raised in this trans digital environment don't bother (and they're in their right to do so) to invest on this behavior.
I may sound like an alien because I'm not a native English speaker and also I tried to sound as neutral as possible lol
oh! I've had the idea of doing this with Wifi signal strength - you could write an app that changes the screen colour based on the signal strength, then do a long exposure shot while moving the phone around. It would require some patience, however!
As a side note: This is why your phone needs to be able to see at least 4 GPS satellites to get an accurate positional reading.
If you were to assume the phone was near ground level wouldn't 3 be enough? Of course you generally can't make that assumption.
@@Sibula what is ground level? When I'm on top of a mountain? Or at sea level? See, that's why it wouldn't work.
I think it does not make a difference how high you are on the ground, the other point of intersection is very very high, about twice the height of the satellite, no?
@@TheXshot we have a height map of the earth, that's not a problem
Your phone uses Assisted GPS (AGPS) - not just satellites, but also cell towers.
I always love stuff involving audio and sound, resonant frequencies, and the fact that we can "extract" frequencies / filter specific frequencies out and stuff is wild. I feel like audio based electronic sensors and such have such a massive future. being able to map / pinpoint the location off sound and how that could progress to audio in 3D space is wild.
Imagine using this camera in a music setting and being able to balance every part of the ensemble. Awesome!
I want to use it to see my farts.
This would work great in orchestral or unamplified acoustic scenarios, but in live amplified performances, which is most concerts, it's better to just use the raw audio from the instruments and then mix and pan that with a recording of the crowd to create a 3D soundscape
@@krikeydial3430 Mine would just be red in the entire room
The way these acoustic cameras work really reminds me of how interferometry radiotelescopes work, i would bet its the same principle (a large array of omnidirectional detectors working together to create a 3d image)
That remindes me of the same thing!
Lol the fact that interferometers are used in cutting edge research and engineering kinda brings this back full circle.
it is exactly the same thing, it's just an acoustic analogy, a lot of the same principles are exchangeable. Another very cool thing that works in the same way is beamforming with loudspeaker arrays but instead of listening in a direction you can project sound to a very narrow direction. this is what most sound bars do and what the leviathan v2 pro does using black magic
This was how I was planning to implement the one I wanted to build. I was told to pick a different project though :(
It's great to hear confirmation that it's the same thing just with a different type of wave.
We use these cameras to find oxygen leaks on airplanes.
When isolating the correct frequency, we can see leaks we wouldn't have found without hours of using bubble solution to many lines and connections or waving special microphones around inside the plane which only work close to the leaks.
When pressurizing planes, these also show the locations of bad seal leaks, and inside electrical panels I've found sources of arcing or they help isolate individual chattering relays within rows of them.
For airplane maintenance, they are fantastic!
While the industrial and appliance applications/overall equipment monitoring capabilities certainly seem useful, when I saw that "eye dropper" tool being used to isolate certain sound sources, I couldn't help but consider the implications for audio production/sound design applications. Would be a dream to have something like this in a recording studio.
My thoughts exactly but applied to programs broadcast on television - isolate certain sound sources - my bugbear is overly emphasised music track competing with dialogue or voice over - I would love to kill the music and just listen to voice track.
And then there are commercials - industry produces cue pips on programmes for breakaway commercial playout which are not incorporated into the tx signal - if it were you could mute top and tail of commercial breaks instead I indulge in timeshift on my PVR.
Why do you film in 60fps in the UK? Shouldn't you want to film in PAL (50fps) to interact better with the electric grid if/when you have any lighting in shot so it doesn't produce a banding effect?
I had so many issues with this when I first moved here, but now I shoot all my videos in 50fps (or 25 for my second channel) and it's grand :)
Yes! But because most displays that my videos will appear on will have a refresh rate that is a multiple of 60, I’ve chosen to film 60 to avoid that judder. I’ve set up all my studio lighting in a way that avoids the 50hz mains issue (my led panels don’t flicker). When I’m filming indoors away from my studio a switch to 50
As a Sound Engineer for music recording studios this must be a huge success (or will be). Making a room "flat" by using material to absorb and/or diffuse sound waves, hence reducing "nodes" - points where reflected sounds of the same frequency collide and become louder and hence create an unbalanced listening space - should now be very easy. Anyone in the biz experienced this yet?
I actually see patterns like the echoes at 9:50 when I’m listening to clapping in rooms with off-white painted cinder block walls, it looks like black and white visual noise as a faint overlay. I always thought I was imagining it, especially since I can only resolve it on certain backgrounds and in rooms made of certain materials (and the room needs to be ~5 to 10m per side). Looking at these though, it seems like, crazy or not, the stuff I’ve been seeing may have some connection to reality.
Sounds like synesthesia!
Tastes like it too!
You’re most likely seeing infrasound! Our eyes resonate just below our hearing range, about 18-19Hz. The wavelength of such a sound in air is about twice the room size you mentioned, so I figure that when you see this, your eyeballs happen to be near an antinode of a standing wave caused by the echoes from opposite walls interfering. And an offwhite cinderblock wall just happens to reflect a lot of both sound and light, so you can better see the visual disturbance caused by the sound deforming your eye. I don’t know why that particular frequency-I assume it has something to do with the acoustic impedance of the vitreous humour-but still, it’s a well documented phenomenon.
I first learned of it in Mary Roach’s book “Spook” where there was a case of a faulty fan in a storage room causing people to see these “ghost” illusions.
It's is observed that many kids and some adults have the ability to see sounds and hear colours
@@kotresh That’s true-and in fact I have synaesthesia-but as a neurological phenomenon, it’s more like “hearing G♯m on a piano _feels like_ seeing something orange” for instance; it doesn’t affect the actual sense. Whereas this has a perfectly good explanation as a physical phenomenon, which anyone could experience, it’s just that it takes very specific conditions to be noticeable.
I sell these cameras (Fluke and FLIR) at work, and it’s truly fascinating to see the problems they can solve in the industry 👍
the fluke versions is built and designed by Benchmark, the basic research was done by fluke.
@@SoYFooD2 its insane how much that company grew in 44 years
The cameras used in the video where made by GFaI in Berlin.
If you'd like to check out a different camera type, take a look at Cavitar and their Cavilux product line. (There may be other products in this space, I've just never seen them). Basically they use a very high intensity illuminator to bring the ambient light level up to things like welding arcs and explosives going off so that a camera can film everything without requiring huge dynamic range. Pretty interesting!
I will, thanks for the heads up.
Oh wow, that‘s a nice surprise. I’ve been following your channel for years now, and totally did not expect to see stuff that I‘m (somewhat) working on. Great video!
Nice pfp
5:42 Nice to see Blender in use! Wow, acoustic cameras are amazing, especially with the point cloud!
Being able to isolate the different sounds and listen to them when you hover is really cool 🤯
Wow! I had no idea this was a thing and now I am amazed. So many uses!
Wow. Never heard of an acoustic camera before, and now that I have.. Of course there's such a thing! It's both a blindingly obvious idea and brilliant in execution.
I remember a Tom Scott video about a Swiss shooting range, where they used microphones to determine how accurate each shot was - basically a simpler version of the same thing - and for some reason I never followed that train of thought to this end result. I guess that's why they make videos and I just comment on them.
Related to shooting as well, there's a microphone array device called a "Boomerang" that can be mounted to military vehicles and is capable of detecting the direction of incoming gunfire.
@@MScotty90 Yea they use the system in large cities as well to help help police with responding to shootings.
Wait til you find out about scent cameras!
Fun fact hyperboloids are also used by TDOA (time difference of arrival) RTLS (Real Time Location Systems) that use UWB (Ultra-Wide Band) RF for location of a beacon. While AOA (angle of arrival) systems use the phase offset at each antenna in an array to determine angle.
Definitely very cool to see what they are doing with microphones!
7:53 This visual gave me the idea of creating a program that tracks a point or a group of points to the ears and connects to another program converting those values into a template for audio. Allowing artists to create "8d" audio much easier, this could possibly change how we experience music and how it is developed, like when the drum machine was created.
I can't believe steve didn't include the audio for him singing with a ukelele in the video 😭
There's actually a reason I didn't! What I was saying while playing the ukelele was about how you can use the camera to isolate just the sound you want (like cropping an image). But we never got that to actually work sadly
I have always wanted one of those. They’re just… well they’re not cheap, and they have to be calibrated often.
Most acoustic cameras today are equipped with MEMS microphones. They do not need to be calibrated often. In fact, factory calibration lasts for years.
That headphone warning did not have enough of a delay.
I was weaking my headset. My wife, in another room, said, "does anyone else hear that awful sound?" I just said, "what sound?" And she said, "never mind, I don't hear it anymore."
As a young person, I felt terrible hearing that high-pitched sound. The speaker did not good despite the fact that I also set it to low volume.
I underestimated just how loud it was going to be. I turned it down, but not nearly enough.
I saw a demonstation of a similar device in an industrial fair like 5 years ago, being an engineer I was thinking to a miriad of possible applications, they told me the bulk of the work they were doing at that time was basically "hole plugger" for like high end cars... being unusually noisy... it helped find the rubber plug someone, for some reason, had forgotten to install in a much shorter time than checking the whole car
I appreciate the fact you were able to explain this in such an easy way to understand.
Hey Steve, great video. This touches upon a product idea that I don't think exists, but I really think it should. For people who ride with a helmet, being able to hear your surroundings is important when riding in traffic, but the sound of the wind will drown out most noises starting at 15mph. I want something that is part noise cancelling headphones, and part hearing aid. I want an array of microphones to create a noise cancelling experience and then filter all of the wind noise out and only allow environmental noise through. To my surprise this still doesn't appear to be a product. I'm under the impression that all the technology to make this product exists, but this product doesn't exist, so maybe I am missing something? i.e. I want headphones/helmet that can give me super hearing.
Hearing protection made for shooting guns have this feature. They amplify ambient sound while suppressing pressure spikes in the audio. Howard-Leight makes a cheap pair that are popular with firearms enthusiasts
a traditional software based sound processor (named DSP, or Digital Signal Processing) subtracts white noise on mic input, but it also subtracts traffic volume, why? because the traffic and white noise overlap in frequency so both is subtracted, so the traffic volume is a fraction of original and has to be amplified to be audible, but the white noise is random, so some noise escape the subtraction and is amplified, creating weird sound. The keyword is "traditional ways", maybe future noise cancelling uses AI for reconstruct traffic sound while leaving out white noise.
@@ayeeniko Thanks for the tip! I looked into their products. Definitely the closest I've seen, but it doesn't look like they specifically do wind.
@@xponen You sound like you have expertise in this area and I value your input. I want to be clear what I mean by traffic sound. If it is a directionless traffic drone noise, subtracting that is fine. I just want to be able to hear signals that have a clear directionality to them, which is where the array of microphones comes in. The nearest microphone will pick up a noise first, but the system filters it by default until the farthest microphone picks up the noise and the system is able to determine that it is the same signal coming from a distinct direction.
Maybe there is also a clever way to do it low-tech. If you design a cleverly shaped helmet or earpiece to stop turbulence from forming maybe?
Tried once Flukes hand-held acoustic camera at technology expo. That was very impressive considering, the whole thing was just a bit larger than regular DSLR
The vacuum cleaner part was probably one of the most fascinating parts of this. I wanna see the other sound peaks
Nothing new here. You can do that with almost any microphone and free audio software Audacity.
These cameras are definitely the future for maintenance and repair. The amount of times a camera like this would have been useful is countless.
This reminds me a bit of when we were installing the subwoofer my husband built. He carried it around to various spots, then we walked around as he manipulated the angle of the speaker cone. We were able to pretty much eliminate null spots that way. We live in a log home, so the acoustics are very, very lively. All the bare wood reflects sound waves quite well.
4:30 I don't think you can actually compare the sample rate to the frame rate, right?
The camera is unique in that for each ""pixel"" (lots of quotation marks) you get a full spectrum, but it would just be like a hyper-spectral camera (cameras where each pixel stores the information of the full spectrum, which for visible light it's even in THz, nothing compared to sound KHz in that sense). I'd love if you could make a video about them btw! :D
The frame rate is something quite unrelated to the sample rate
Similar to the JWST which takes a picture at every frequency or a spectrum at every pixel, depending on how you want to look at it, same net result.
Correct. The Fourier transform shows how the location over time and the frequency are related. The range of frequencies you can capture depends on the sampling rate (Nyquist-Shannon sampling theorem) and the resolution of those frequencies depends on the number of samples you take. Two samples back-to-back tells you how the air pressure is changing at each microphone, but it doesn't give you a frequency, since that's a property of a waveform over time. Only with many samples can you separate out those variations into frequencies.
I was intrigued at this technology but I burst out laughing at 1:57 because it looks like a shitpost
1:00 We want to hear the actual song!
I remember in 3rd grade, my best friend's dad told me about an idea for an invention. Glasses that you can put on your face and see sound as a visual. He said it would have all kinds of colours for different frequencies. It's crazy to think that by the end of my lifetime, I might actually see that entire technology develop to his seemingly impossible idea from 2005.
"cones of confusion" is a really good description of my perception and understanding of the video :D
If anyone finds it difficult to see why you need 4 mics to locate the sound, think of it this way:
Imagine 2 intersecting spheres, the intersection they make is a ring. Now push that ring into a 3rd sphere, and the ring will contact the 3rd sphere's surface at 2 points still, so you need a 4th sphere.
Or you can think of it like triangulation
I understand the cones and spheres are more accurate to how it actually works, but understanding it as distances between points describes it identically, without the waffle.
I'm acoustic, I should be able to see sound
internet has ruined the word "acoustic" for me
@@mattmartin7028 fr
Just a note, at 0:40 when you're talking about Automatic camera exposure, you've set it to Aperture mode, not Automatic. Being unfamiliar with that particular camera, I THINK if it has an automatic mode, it would be two more clicks, just after P (for Program) the little A inside of camera symbol.
On a digital camera, it could be argued that automatic exposure would be relevant on S (shutter) A (aperture) or P (program) since the camera automatically adjusts settings on those modes to get exposure correct. And automatic ISO could even apply to M (manual) mode as well!
But the 'A for Automatic' that's being insinuated isn't entirely correct :]
This reminds me of when I was a bicycle mechanic and would listen to customers rolling their bikes into the shop and use the sound in my diagnosis. I can still diagnose passing bikes which sometimes gets annoying.
in 4:02 , you can see steve suffering the horrible high-pitched torture. and still see signs of pain when the camera goes acoustic mode.... for your dedication to science i solute you...
6:56 so there is a advantage for anime girls with 2 more cat ears.
Interesting use of the acoustic camera to detect faulty wheels on trains. Due to the unpleasantness in Ohio, we now know that thermal cameras are often used for the same purpose (though they more often detect brake faults).
Related note: if your train wheel is on fire, it might be time to service the brakes.
The colloquial name for those detectors is "hotbox detector." That name dates from the days when axle bearings were something like a Babbitt bearing encased in a "box" packed with oily rags. When the bearings wore out, they could very easily catch fire and, as many freight cars still had lots of wood in their construction, that could be quite catastrophic.
These days, roller bearings are required for cars used in interchange, so they wear out less frequently and there's not quite so much flammable material nearby when they do, but as we've seen, it still happens. And, as you say, they also detect stuck brakes.
They're mostly not thermal cameras, though - the size and construction of a wheel doesn't change much, and the train itself provides the necessary scanning motion, so a simple passive IR sensor pointed at the spot where an axle goes by is sufficient. Because they can be so simple, they've actually been around for a long time - Wikipedia says since the 1940s.
@@RonParker interesting. Do you know if there were thermal cameras in use on the Norfolk train? I could have sworn I saw the actual footage - or at least the temperature readings - from the train. One reading shows it elevated, but not enough to trigger an alarm, and by the next reading the thing was already energetically on fire.
"If you have your camera set to automatic exposure" *Sets the camera to aperture priority* Eh, close enough!
I'm researching into echolocation and I feel like this video's gonna be really helpful for it
I went to an art installation in Vienna once that had fixed piezo sensors in a room to achieve the same effect. They had video projectors projecting lines onto the edges to features in the room, and as you made a noise they would ripple the walls etc. in that section of the room. Very eerie!
Bro can see sounds??????
Who can afford Photoshop? Can you make it available elsewhere? You know that's affordable.
I cannot even begin to tell you how much I appreciated that headphone warning, thank you!
1:52 i hate that you have to hold the button when using the dyson. idk if they know but toggle buttons are a thing and they are easier for my index finger muscle
@Steve Mould about what you said about the vibrating aluminium bar/rod. Although it's possible for the vibration to be longitudinal the fact that the sound originates at the bar/rod ends does NOT imply that the vibrations are axial. Since the rod is constrained at its midpoint (acting like a node) and not at the 2 ends it can not behave like a guitar string(what you show is the 1st mode for a bar pinned at both ends), but can still vibrate in flection with the first bending mode having the largest displacement at the ends (like each prong of a diapason). That said the bar was behaving one way nor the other? A reasonable assumption is usually that the mode with the lowest natural frequency is the main one(and usually bending modes are always lower than axial because usually axial rigidity is higher) but given the way you excite the bar ("pinching" on both sides) it's possible that the bending mode is not excited at all, while the axial is. Ho To tell the 2 possibilities apart? First thing that comes to mind for someone who was taught how to without this amazing equipment would be "Frequency" : figure out with formulas the frequency for each possibility and confront with the measured frequencies. But since we have this "sound video" the source for bending mode would be more distributed along the bar with the max at the ends gradually getting less and intense towards the constraint. Axial instead originates only at the 2 cylinder bases(apparently what we see in the clip).
Also Intuitively: since I mentioned diapason...this bar is aluminium and despite having a larger diameter than the usual diapason prong I assume it's far less rigid in bending than a steel diapason prong, and the length is likely 3or4 times one of a standard diapason (that is already quite high pitched) longer length and lower rigidity bring the natural frequency down so there is no way this bar would produce so much higher pitch sound than a standard diapason
Phased arrays and wave propagation greens functions in action. I absolutely love it!
Whenever someone asks me "what kind of super power would you like to have?" I'd always say that I would love the ability to see sound. Like think about everything around us that makes sound, from our fingers tapping onto a keyboard or even the sound of clothes rubbing together as we walk. Just so much stuff producing sound all around us!
I love that the video started with a clue to flaws in cameras
We use the FLIR acoustic camera to identify leaks in compressed air systems. What is great is not only does it locate the source of the leak, but software can also identify the volume of the leak. So after mapping an air network we can say for example "you are spending $50k per year in air leaks, they need to be fixed". Compressed air is very expensive to make at an industrial scale and air leaks are a big issue in large buildings.
No 4 minute sponsor, showed us what we wanted to see at the very beginning of the video. Thank you
05:32 "The hyperboloid of confusion." There's a reason why dogs cock their heads when they hear something interesting.
Timing is one thing, but the change in db is also an effect which can help us decide if the sound is near or far (if the change in db is big or small between our ears).
There is probably more information available to our ears as well, since our ears are not uniform, meaning a sound actually going "behind" has a different curve then a sound moving "infront" of the ear or if it is.
This was the one. This was the video that finally got me to sign up with incogni
9:15 oh my god these are things I always wanted! Especially the echo purely as a single pulse traveling and reflecting around. Insane!
thanks for giving me a headphone warning 0.3 seconds before you made the sound. Really gave me ample time to adjust my volume
This is juicy information
@Steve Mould, the ear isn't a simple analogue-to-digital converter, so the brain doesn't receive waveforms to analyse.
The cochlea performs spectral analysis, the output of which is conveyed to the brain.
Hi, great video! I’m lucky to work with acoustic cameras on a daily basis. I live in Poland where people are price sensitive but thankfully there are acoustic cameras like CAE SoundCam and others which are affordable, have high quality and performance. I can enjoy working with them everyday. It's the really helpful technology for engineers!
Captured or not, sound is so beautiful! Thanks for the video, it is cool!
Indeed this is one of those topics I've been researching for quite some time.
I worked on these issues with analog technology more than 30 years ago , the renderings were vector graphic monochrome however my focus was on AI sensory integration for environment object characterization .
I saw this idea on a website that collected weird ideas many many years ago, & I’ve always wondered why this great invention hadn’t been created ‘yet’ ?
The application that I was thinking, would be absolutely fabulous, for this would be to put on a balloon and float it over a forested area and listen for big foots .
Your explanation talent and your ability to make me exited for anything is so remarkable. You could make me excited about wood sticks.
We have these cameras at my work. One of their uses is to find steam leaks as superheated steam is often invisible
4:30 The time-domain sample rate can be 48, 96, 192kHz or what have you, but the live processing is almost certainly not being done at the same rate. They're typically done on "blocks" or "frames" of samples which are buffered up. If you wanted to process a full frame for every sample coming in you could. You would end up with a ton of frame overlap and it would be crazy inefficient and probably unnecessary.
As a synaesthete tortured by constant noise, i need this acoustic camera. 6:56, finally! This is exactly what I need, a device that can visually indicate the location of a sound source.
5:39 hyperboloid of confusion does not cause a problem for careful brains because the moment you enter an environment, your brain creates an audio-3d map of the enivronment using how which area of the room changes the sound. let me clarify; for ex: rear side of the room amplifies lower frequencies and upper area has less echo or front of the room amplifies the high frequencies etc. since your brain picks this up instantly, when you move the sound around, you can locate it not only by amplitude but also using frequency analysis. this is also how proper surround mixing to stereo is done, (only by experts) forget only playing aournd with amplitudes or phase differences. the whole tone should show a difference also not only in frequency but in time base as well, as in some frequencies takes longer to diminish in an environment due to the that environments natural resonances.
2:09 Vacuum cleaner manufacturers do NOT want a quiet vacuum because people won't buy it. Customer assume the noisier one has more power.
5:00 I always felt I got direction from sound by the varied intensity between my ears rather than the delay
The shape of the ear actually helps at determining the direction of the sound source.
Really interesting video, thanks!
2:43 railway research: acoustic defect detector to prevent catastrophic failures 9:23 room "echo" technically reverb(eration); echo applies to canyons and such
6:58 Trinnov AV receiver uses the similar technology to identify the speaker location in 3d space and optimise the sound in real time while other receivers simply can't . I didn't know these techniques can be used for other purposes than a theatre. Amazing video👍🏾
Fascinating tech. I had a lot of fun watching this Video. Excellent presentation.
Acoustic cameras are really cool! We got a couple of setups at my university.
The analysis methods using crosscorrelation are already pretty interesting, but there are methods to get much more resolution out of the array.
You can use inverse convolution with the point spread function of the array to reduce the interferences at high frequencies and also get better results for lower frequencies.
They were also used to compare the noise made by flying owls and other birds, to see which mechanism is responsible for these noises and so on.
that is really state of the art technology right there.
the metal rod sound is the first time I've ever turned down my volume due to sheer panic. at first when you said headphone warning I was like: "nah it's probalby not that bad" but then when you did it I jumped in my chair and instantly turn down the volume all the way XD
this is really cool! I bet there's a lot of complex logic to interpret these signals. Because the speed of sound is, well, finite, and the camera is super slowmo, when the camera registers something it has already happened a while ago. This is specially relevant when looking at those echo blotches.
Hey. I´m from germany and i love your videos and your way of explaining complicated facts of science that i´m very interested in. Thank you.
I HAVE WANTED ONE FOR YEARS THANK YOU
What's amazing is that our human ears are able to pick out direction, elevation, frequency, volume and motion (doppler shift)... with only 2 ears. What's really amazing is how much better Owls, Whales, etc, are with their sense of hearing. We need a battery of microphones and computer software to try and come close to replicating what they can do naturally.