Transforming my studio with AI Designed, 3D Printed Acoustic Panels
ฝัง
- เผยแพร่เมื่อ 8 พ.ค. 2024
- In this video, I unveil a groundbreaking transformation of my workshop's acoustics and aesthetics using AI-generated, 3D-printed Acoustic QRD panels.
If you're into DIY, AI, 3D printing, or acoustics, you won't want to miss how I combined technology and creativity to create a unique acoustic solution for my space.
To find your own panels, visit www.markrhodesmaker.com
Song: Savfk - Journey To The Stars
License: Creative Commons (CC BY 3.0) creativecommons.org/licenses/...
Music powered by BreakingCopyright: breakingcopyright.com
Song: TheFatRat - Time Lapse
License: Creative Commons (CC BY 3.0) creativecommons.org/licenses/...
Music powered by BreakingCopyright: breakingcopyright.com - วิทยาศาสตร์และเทคโนโลยี
Hi, I'm an acoustic engineer, and these diffusers will only affect high frequencies. For a diffuser to scatter sound, the individual elements of the diffuser (in this case, the individual squares/cubes) have to be at least the same size as the wavelength you're trying to scatter (preferably larger). From your website I see that the largest heights you offer is 50mm, and it looks like the edges of the squares are also around that size. So if I'm assuming cubes with 50mm on each side, (343 m/s) ÷ (0.05 m) = 6860 Hz, meaning any wave under 6860 Hz will not be affected by the diffusers. In a logarithmic scale (the one relevant to the human ear), the effective range is about the highest quarter of the audible frequency range
So how could that be improved, with smaller squares, larger ones, or a fractal like Perlin noise? Perhaps using a material that is softer or porous?
@@johnvonachen1672 Both larger elements and softer material help, though it may be hard to achieve in the same material (impossible in 3D printing). You could use a combination of some absorptive panels + some diffusers. Absorption just helps all around to acoustics mainly controlling the reverberation time. Scattering/diffusion is mainly to break resonant modes and make the sound field more diffuse (meaning sound behaves more consistently throughout the room). Each ones do their own job best, and give best results in combination.
If you look at anechoic chambers, they have an absorptive but spike-shaped material, which reflects the sound onto other spikes and so makes the sound get absorbed multiple times on the wall. But you will also see that the size of these spikes are quite large
Thanks for taking the time for the well thought out reply, and I really appreciate you bringing the maths too! I have some follow up questions, why the speed of sound in your formula? What size do you think would be best for spoken word? Would the top surface dimensions (x/y) matter as much as the z dimension? Thank you!
@@MarkRhodesMaker Hello! The formula for speed of sound is (Speed of sound) = (wavelength)*(frequency), meaning that you can get the wavelength by dividing the speed of sound over the frequency. So here I worked backwards, starting from the size of the elements (in 1 dimension), then use that as a wavelength (meaning, the largest that can be diffused), and then finding out what frequency that wavelength corresponds to.
For spoken word, typically you have _fundamental_ frequencies between 100-240 Hz including both men and women. For, let's say, 150 Hz the wavelength (and therefore, the minimum size of scattering elements) would be 2.28 meters, so that's quite unfeasible. You could instead choose to look at the whole frequency spectrum of voice and control the strongest part of the spectrum, and there you typically have the most power between 800 and 1600 Hz. For 1000 Hz the wavelength is 34cm, which is still considerably large for scattering elements but not as physically impossible. For speech you might as well instead control it with absorbers instead of diffusers. Membrane absorbers are the type that are typically used for low-mid frequency range, while porous absorbers (the "foamy" type stuff) are more for mid-high frequency range.
I should also clarify that these frequency limits I'm calculating are not 100% accurate but are just rough (but pretty close) estimations of what the effective frequency range is. It is also not a hard-cut yes/no limit but actually a gradual transition. Still, this rough calculation gives you a pretty good idea of the frequency range.
I would say that all 3 dimensions are about as important. This I'm not 100% sure but probably the Z axis/height is just slightly more important. But XY dimensions still matter considerably.
I hope you find this information useful!
Even without the math it's pretty obviously snake oil. Hard flat plastic surfaces do poorly as acoustic treatment.
Interesting project - thank you.
Feedback: the audio on this video is really poor in places (ironic, I know) such that it's near impossible to understand a few of the words you're saying.
Your speech appears to be being clipped too much in places - a bit like when you have a bad "line" on a mobile phone call. Perhaps over-compressed or gated audio? (I'm no expert though)
Thanks all the same, still an interesting video.
Oh that is very ironic! OK, that's good to know. I'm still getting used to my new recording kit, and I'll make efforts to make sure it's recorded a bit clearer in the future. Thanks Chris!
Seems like a lot of thought went into making the panels, but not where to put them? Not sure you are going to get much acoustic bang for your buck under and behind those cabinets. Cool project though!
Honestly, that's a REALLY good point and it's not lost on me. The best place for the panels would actually be on the roof above me, and also the cabinet doors behind me. Visually it's in the best place, acousticly? Not so much! Thanks :)
Yeah, I agree. Acoustics aside, it makes for a very cool looking backsplash....In white and fairly shallow height I think it would fit nicely in a modern kitchen.👍
@@ThePhilboxDon't say that, my partner watches the comments and wants a new kitchen!
The panels looks amazing!
Thanks very much!
Man I loved being part of the brain trust as this came together! Just wish I could have been down there with you to take some acoustic measurements before and after!!!
Next time! Maybe if you moved a little closer to Australia? :)
Are skyline diffusers limited to the block shape for each well or can they be, say… hexagons?
Because we know they are the bestagons 😁 and would match my plans…
I've been wondering the same thing myself. There is no reason that they should not work and theory says "yes, this should work" however real measurement would have to be done on how well it scatters sound compared to a square
Cool looking panels! Not a huge deal, but I noticed what looks to be constant camera lens focus breathing on you wide shots. it's most noticeable when looking at the color changing acrylic sign.
Thanks! Yep I noticed it too during editing, but by that point it was too late! Next time I'll go with manual focus instead of face detect :)
11:37 something interesting to look into regarding this one would be foaming filament. Normally it's mainly used to print wings or such for RC airplanes and the like.
But if there happens to be foaming TPU available too, then that's basically 3D printing foam.
Interesting idea, is there any reason or application in particular you are thinking of?
@@MarkRhodesMaker not really, I was just wondering if it'd make a noticeable/measureable difference if the same panels you have were printed in a "foamy" or "softer" material.
As in, would the absorbtion add into the diffusion or not.
@@BloodyMobileI suspect it would not be a cumulative effect unfortunately.
Spider hatching cubby-holes! Vacuum often.
I'm not going to be able to unsee this...
haha! Fun video. Thx! Extra care as an Aussie when it comes to spiders.
@@MarkRhodesMaker
Love the idea, great work.
For the prompt, you can change it slightly.
You are a mechanical engineer with 20 years of experience. You are specialized in openscad.
Your task is to generate an openscad model that approximates a QRD diffuser for 3d printing.
I will give you more instructions for the parameters.
Think thru you're answers before responding.
You could even started by adding before a snipped of qrd from the pdf as an example.
Best results using GPT 4.
But the prompting change the output significantly.
// Parameters (to be customized based on your instructions)
N = 7; // Prime number
W = 10; // Well width in mm
Dmax = 50; // Maximum depth in mm
H = 100; // Height of the diffuser in mm
totalWidth = N * W; // Total width of the diffuser
// Function to calculate depth based on quadratic residue sequence
function calcDepth(i, N, Dmax) = Dmax * (i^2 % N) / (N-1);
module well(depth, width, height) {
translate([0, 0, height - depth])
cube([width, width, depth], true);
}
module qrdDiffuser(N, W, Dmax, H) {
for (i = [0 : N-1]) {
wellDepth = calcDepth(i, N, Dmax);
translate([i * W, 0, 0])
well(wellDepth, W, H);
}
}
// Generate the QRD diffuser
qrdDiffuser(N, W, Dmax, H);
Thank you! Great idea to re-engineer the prompt. It's something I often forget to do
I tried that code, it didn't work! It came out looking like a space invader. But pretty cool regardless to see my video motivated you to try generating some code in GPT :)
Fantastic idea, they look, and from your example, sound, great.
Thank you! Cheers!
Yeah, chat gpt isn't to be trusted AT ALL with that kind of specialized equipment.
Looks good tho.
Definitely not without supervision!
Lol, dude, what? This is great. (In a not good way).
Thanks? I think?
As someone already wrote above, these panels are completely useless for frequencies below around 6.5k Hz (so basically everything that is boomey and especially spoken word). Don't fall for it...
Thanks for taking the time to reply! I'm a guy in a workshop making what I thought was a cool project with a 3D printer. What is there to fall for? :)
@@MarkRhodesMaker For instance buying your designs that will not produce the advertised result?
I have nothing against sharing work and research, but we already know how acoustics work. And your designs won’t work for frequencies below roughly 6.5kHz. It’s physically not possible.