3D-printed device splits white noise into an acoustic rainbow without power

I love corregated concrete! Those highly ordered early reflections can make a cool sound if you stand about a few feet away and clap your hands. I've wondered if it was possible to vary the bands to encode a designed sound in the reflection. Probably.

That is a very wild shape indeed. I wonder if there is some analytical way to derive these or if some type of search algorithm will remain the best way.

Thats crazy ! Thanks for sharing

It's not really machine learning, it's just optimization (optimize the space of shapes to create the desired filter). The paper doesn't call it ML either

Cool, but I'd rather see a nicer shape, like a human cochlea, which also splits sounds into frequencies.

Perhaps if they added some more constraints (e.g. on smoothness) they would end up with a shape like that.

What sort of ML method would do shape optimization?

Reminds me of Ben Underwood, the blind kid who used echo location around the house, playing basketball, riding his bike around the neighborhood, etc: https://www.youtube.com/watch?v=fnH7AIwhpik https://en.wikipedia.org/wiki/Human_echolocation#Ben_Underwo...

Before practical applications, my first thought was a grade-school science fair entry. It’s novel enough for judges not to have seen it for a couple of years.

I’m game to do some heavy lifting if you’re serious.

I’m wondering if you can change the shape in such a way that rotating one would produce an arpeggio.

I'm in.

Although I think this also has applications as a mechanical control mechanism, replacing expensive potentiometers with relatively cheaper 3D-printed parts. I've dreamed-up an optical/laser variant in my head a hundred times, fun to look at it in the audio domain.

I can't find it now, but there was a youtube video of a guy building a fence to block road noise by building interlocking pottery 'bricks' with cavities that would absorb certain frequencies

This is the first device of that kind. It's a research paper. The method described works. They weren't going for the limits of the technology here, but proving that it works. Others with different requirements of frequency ranges can create their own versions.

Terrible! I was expecting at least a 7599-13601 Hz split.

Ah, so you're pointing out that you could scale it up and down and stack them to convert white noise into chords? Cool!

I think the ear relies more on resonance (unrolled it's like a stretched rubber triangle). The higher you go, the more it reacts on higher frequencies. This one here seems to just use passive reflection to get the effect.

There are ways of telling if she is a witch

Do you mean a prism?

It would need to do something like spit out each frequency in a different direction then you use light material in circle that flaps when sound energy is transmitted in that direction. Sounds possible. Analog computer of sorts.

1. Start with a non-crazy shape that sort of does the same thing but worse (ie non-linearly or with big power fluctuations for different frequencies). Almost all shapes will interact with a sound wave in a frequency dependent manner so you can quickly scan through many initial options.

2. Define a goal function describing the desired outcome (like the "rainbow" shape of the linked article)

3. Permute the shape you started with iteratively with algorithms like simulated annealing, using the goal function from 2 as a means of defining the quality of the current solution.

The actual scientific paper (https://www.science.org/doi/10.1126/sciadv.ads7497) is worth reading. The goal function is the "Figure Of Merit" (FOM), described in equation 1. They also make a two-prong version they call the "lambda emitter" that takes in a mix of sound frequencies and directs low and high-frequency sound waves in different directions.

There are various algorithms for finding approximate optima to messy fitness landscapes.

Genetic algorithm would be one approach. You map the printable shape space onto a bit vector and define a fitness function (optimization objective function) to guide the search. Hardest part is coming up with a fitness function that gives you what you want, this gets tricky when you are trying to balance multiple constraints.

I mean we already built the pyramids...

Just like it says in TFA. Go figure.

But for sounds