When it’s time to put together the annual Christmas card, most families take a few pictures of the kids, slap on a generic greeting, and call it a day. It used to be fairly common for the whole family to get dressed up and pose for a special Christmas picture, but who has the time anymore? It’s not like we have hours and hours to slave over a unique and memorable gift we can mail out to a dozen (or more) people.
Unless you’re [Chris Wolsey], anyway. Rather than take the easy way out and simply mailing some pictures of his daughter out to friends and family, he recorded her giving a Christmas greeting and turned the waveform of her voice into a framed physical memento. Way to wreck the curve for the rest of us, [Chris].
As it turns out, getting sound into CAD software isn’t exactly straightforward. To start, he made a recording of his daughter saying the words “Happy Christmas From the Wolsey Family” with Audacity, and then took a screenshot of the resulting waveform. This screenshot was then brought into Adobe Illustrator and exported to SVG, which Fusion 360 (and most other CAD packages) is able to import.
Now that the wave was in Fusion 360 he could scale it to a reasonable size, and use the revolve function to bring it into three dimensions. Cutting that object in half down the length then gave [Chris] a shape which should, theoretically, be printable on his FDM printers. But unfortunately, it wasn’t so easy. His personal Anet A8 had a tough time printing it, and the Prusa i3 MK2 at work didn’t fare much better. In the end, he had to make the leap to SLA, getting the shape printed on a Form 2 via 3D Hubs.
With the finalized shape in hand, [Chris] just need to put them into production. Printing them all via 3D Hubs wasn’t really an option, so he decided to make a mold and cast them in resin. He printed up a mold box, and after fiddling around with the mix a bit, was able to settle on a resin which allowed him to de-mold the shapes just 30 minutes after pouring.
Finally, he made frames for each cast waveform, and printed up a little label explaining just what the recipient was looking at; even going as far as showing which word corresponded to which section of the shape.
This is a fantastically executed and documented project, and while it’s too late to whip up your own version this year, we have no doubt they’ll be a few people “borrowing” this idea next time the holidays roll around.
It’s never too early to start planning for next Christmas. We’ve covered unique takes on the traditional holiday card before, as well as a sleighful of holiday decorating projects.
I’d tried and failed at a similar attempt. What I did was record a sample, get an FFT for a bunch of narrow bins, and imported the resulting intensity over frequency and time grid as a height map in openscad, but the results were largely unprintable as many of the spikes we’re narrow enough in either frequency or time to be brittle or for the minimum printable size to not slur the sounds too much.
I didn’t, of course, think to farm it out to an SLA and cast from the resulting high quality print… Way to go [Chris]!
I want to see the 3D printed machine to play the wave back.
You wouldn’t be able to, there’s nowhere *near* enough detail in that. All that’s being recorded in that shape really is the envelope of the volume at each point in time. There are billions of possible recordings that could fit that envelope.
Are you sure? https://en.wikipedia.org/wiki/Optical_sound
Yes, sure.
Magnitude information only, with no phase information, it would sound odd.
Sorry I was thinking FFT, the info is in the time domain, but yes low resolution. There is software that will convert images into sound, but they assume the information is magnitude only (and the phase of everything is zero at a guess)
https://www.youtube.com/watch?v=S64FROErFYA
Play it back with a laser or even easier off the shelf, a leap.
A leap would certainly have the resolution.
Next year you can send them something that lets them play back the previous years greeting.
Maybe next year he could send a plaque with ones and zeros encoding the audio in low bit rate audio codec frames using codec2. 700 bits per second should give pretty good fidelity.
Pretty neat thing to do. All that research and effort, only to leave the e off silicone.
I have to wonder if it was a simple typo or if he’s one of those that doesn’t know the difference between silicon and silicone. If I ever invent a time machine I’m going to go back and change the name of one or the other to put an end to this problem.
Just you wait till they start hiding spy equipment in them. You got your silicon in my silicone.
The result looks more like a thing you put in the top of a Christmas tree than a waveform.
Very nice looking final product, but have to say this strikes me as something better left for a parent to make for their own child. If somebody sent me this, I’m not sure what I’d really do with it.
i get it, but Grandma woulda grabbed that and whacked me over the head with it.