letterpress record process documentation / blog, 2017

test #7
This is exciting! This test worked pretty perfectly. I ran it on a number of paper types but the one that worked best was actually an uncoated Xerox paper that Alex at Paperworks recommended (need to go back and double check the name, but it was suuuper cheap, which rules). Amanda Ghassaei and I had an email correspondence about the RPM issue in her code, which she recommended a fix to. And it was fixed indeed! So this record runs at 45RPM and stores about a minute and a half of sound. The sound degrades as you move closer to the center (like any record) and there’s still a lot of fiber noise, to be sure, but you can hear a song fully. Next steps are to see how much larger I can make the amplitude of the waves before the needle begins to skip, and to see if there are ways to diminish the noise of the paper fibers. More to come soon.


test #s 4, 5, & 6

I met with Dan Wood at DWRI to discuss the project and he was incredibly kind and helpful! We ran test #3 through the big press at Dan’s shop, and through that process I discovered some new paper types that work. There was one, almost like construction paper, that held the needle in the groove really well. As the audio became clearer to hear, it became evident that something in the code produced an error in the playback speed of the record. The 78RPM disc didn’t quite play at 78RPM, and the 45RPM disc didn’t quite play at 45RPM. I emailed Amanda Ghassaei, who generously replied noting that an error in her code had been causing playback speed inconsistencies, but that she hadn’t found time to fix it. While in Los Angeles for a week, I visited my friend Lukas, whose engineering curiosity and mathematical background were extremely appreciated. We looked through the code together and found that the playback speed errors have to do with the sampling rate of the incoming file vs. the outgoing audio. We tinkered around and I sent a new plate to press, which didn’t *quite* solve the issue, but it’s closer! And you can really hear the audio! Also big thanks to Alex at Paperworks for giving me some new paper to test things out on and Rebecca at Boxcar Press for helping me make letterpress plates out of extremely unorthodox and data-heavy Illustrator files. Once the playback issue is solved, I’ll be going back to Dan’s shop to do a huge test run on a bunch of different paper types using the commercial press!


test #3

I tested the same plate as test #2, but this time on Neenah’s Plike paper. It worked so well! The record plays and sound comes through clearly, though th needle still skips. Feels like I’ve eliminated two of the three main variables of the project...

1) Generating the plate image so that it actually plays sound back
2) Finding the right paper type that offers good depth of impression and isn’t so fibrous that it just sounds like paper scratching itself but also is rigid enough to hold up to a pressing.
3) Getting a deep enough impression so that the needle stays fully in the groove.

I’ll be meeting with Dan Wood at DWRI next week to explore printing a record on a clamshell press, which should offer a more uniform impression than the flatbed press at AS220.

test #2


The plate worked really well! Underneath the fuzz of the paper fiber there is definitely music (harder to hear in the video because of phone audio, and also because it was around the tenth time I’d played this record so it was getting kind of worn out.) Having trouble with the paper type - a Xerox coated cover weight paper - because while it isn’t as fibrous as other papers I’ve tested with in the past, it also doesn’t take an impression very well and the needle tends to skip out of the groove. I’ll be testing this same plate next on Savoy (a very fibrous soft cotton fancy paper usually used for wedding invites) and Plike (a Neenah paper that’s kind of a wild card - soft, like cotton papers, but designed to be ‘plastic-like’ in its texture). 


test #1, 3/6
Starting back up after months away from platemaking under funding from RISCA’s Design Innovation Grant. I knew that the plates I’d made over the summer worked best with simple sonic information – a sine wave was the only fully successful test. I switched computer programs from Mitsuhito Ando’s CuttingRecordGenerator to Amanda Ghassaei’s LaserCutRecord script, which first generates a text file of coordinate points and then plots those points as a waveform. This allowed me to achieve a higher fidelity on the image, because the points could be plotted more easily at a larger width. The sketches for laser cutter output at a .001pt width, and the minimum plate width for the letterpress is .25pt, so there’s scaling up that needs to happen. Previously, the scaling-up had resulted in lots of clipping –– the lines would collapse into each other as they were enlarged. This plate, while unsuccessful, was the first step towards a cleaner line output. Here are images of the parameters and the output.





how to make a letterpress record
1) Process a digital soundfile in Audacity with an inverse RIAA curve.
2) Run the file through Amanda Ghasessi’s LaserCutRecord Processing script. I’ve made modifications to this script to get lines at the proper width which I have published on GitHub. More info on steps 1 + 2 from Amanda Ghassaei’s Insctructable for laser cutting records.
3) Make a negative of the image and expose it to a polymer plate. I do this by sending the file in to Boxcar Press in Syracuse, NY.
4) Put the plate on the letterpress and run paper through, experimenting with the number of extra sheets as carriers to increase the pressure.
5) Play the record!