3D Printing Records

3D Printed Record

This is a working record created with a 3D printer. [Amanda] came up with a process that converts audio files into 3D models. These models can be printed and played on a standard record player.

The real work is done by a Processing sketch that creates a STL file. [Amanda] started off by trying to create a sine wave. She used this test to optimize the printing process. Then she used Python to extract audio data from WAV files and modified the processing script to process the data. After more tweaking, she was able to get a reasonable signal to noise ratio and minimize distortion.

The resulting records have a sample rate of 11 kHz and 5-6 bit resolution. The sound quality isn’t going to be the same as commercially pressed vinyl, but you can still make out the song.

Objet Connex 500 was used to print the records. This UV printer has a 600 dpi resolution, which is means it’s more accurate than extrusion printers. Your mileage may vary using different printers, but all of the Processing and Python code is available with the project write up.

After the break, watch [Amanda] spin some 3D printed records.

52 thoughts on “3D Printing Records

    1. Sure they’re intelligible. He’s singing: Here we areg nowg enter hey us were her gerg nu an her gug hu etc. I like the song tho.

    1. It’s not truly analog. It has a limited resolution as a result of the UV printer, so you’re not getting a truly analog and continuous waveform when printing – it will have discrete “pixels” that make up the waveform.

      1. It *is* analog. Everything is analog, including digital.

        Digital audio is a form of analog where the BCD values are analogous to the acoustic pressure and is limited in resolution by the noise of the playback DAC jumping from value to value and those values are the result of unavoidable storage space limitations.

        Vinyl audio is a form of analog where the offset of a track is analogous to the acoustic pressure and is limited in resolution by the noise of the playback needle jumping from granule to granule and those vinyl granules are the result of unavoidable manufacturing limitations.

        3D-printed record is a form of analog where the offset of a track is analogous to the acoustic pressure and is limited in resolution by the noise of the playback needle jumping from granule to granule and those plastic granules are the result of unavoidable printing limitations.

        It just so happens that the limitations are worse on the last than on the former, at least with this printer/technique.

        All forms of analog have limitations and those limitations can always be at least partially expressed in terms of frequency ceiling and noise floor, in the case of the later it doesn’t matter wether you’re talking about bits of information or bits of material (vinyl).

        1. Everything is analog at the point where it’s expressed as an analog signal (like sound pressure). What’s your point?

          You actually seem to be trying to argue that everything is *digital* by talking about the “resolution” of the analog playback of digital signals. Truly analog signals don’t have a resolution because they aren’t quantized.

          The fact that one signal is analogous in some way to another signal isn’t the modern definition of “analog”, though I suspect you know that.

          So digital audio isn’t a “form of analog” – it’s a quantized (i.e., digital) representation of an analog signal. Not to mention, the sample resolution and the playback resolution can be very different.

          1. “don’t have a resolution because they aren’t quantized”

            This is where you’re wrong. Yes they are. Look at any analog medium under a microscope. Even the original acoustic signals are quantized at the end of the day – quantum physics.

            “You actually seem to be trying to argue that everything is *digital*”

            Well, everything can be expressed with digits. If you have proof otherwise a lot of mathematicians would like to hear from you. That said “quantized” is not the same thing as “digital”, everything that is digital is quantized but not everything that is quantized is digital simply because there are things that nobody has bothered to count and write down in digits (an ADC’s job).

            “the sample resolution and the playback resolution can be very different”

            You mean just like the vinyl turntable can be of lower quality than the press the vinyl was made on?

  1. Great now the RIAA is going to be all over 3D printing. Demanding tax revenue per 3D printer sold, Just like blank CDs. Because you can’t prove everyone isn’t going to go out and spend $40K+ to make low quality LPs, or some other shifty logic ~.~

  2. sweet now print a phonograph cylinder the same way. I don’t have a 3d printer so I must make requests to live vicariously through others.

  3. I have to wonder what the impact of the (albeit 600dpi) ‘steps’ are on the stylus – can’t imagine it’s good for it.

    1. Assuming that either both materials are of comparable hardness or the difference is negligible relative to the difference against the metal of the stylus, the impact on the stylus is actually no different than the impact of a true vinyl containing a recording of this noise.

      Of course those assumptions could be wrong, I’m not a materials guy, but this should do it until a materials guy chimes in. :-)

  4. I hear some kind of sound below all the audio and when there is no music playing. It sounds almost like servo movements. Where is this signal coming from? It’s pretty consistent which is why it’s interesting to me. It’s not just speckled dust noise or printing resolution issues. It’s a consistent “wow wow wow” sound.

  5. Nice approach… the artifacts caused by the cartesian stepper path should vanish when a rotary table is used. Then, things will start getting really interesting.

    1. damn, I missed the description. No steppers here.. however, it’s still an aliasing artifact caused by drawing a circle rectangular coordinates.

  6. That’s funny because I was going to try to scan vinyl records in order to digitize them. Not by putting them on a large scanner, mind you, but rather using a CCD from a scanner as the “needle”. You know, (half) a rotation, some processing of the image, done. Easy. ;)

    And then, by printing and scanning and printing and scanning and printing records, we could see how the quality of the audio detoriates like this one project video on Youtube where someone re-encoded audio/video content via some other method which escapes me right now…

    But seriously, I was really going to scan vinyl records…

    1. It surely will NOT be easy but this project has come to my mind every now and then. We will first have to figure out how to properly image the grooves.

  7. s once the process gets more refined and on track, you should be able to print your cue marks, have multiple loops in a row for scratching. I really dig the “WOW” sound that happens with each revolution.

  8. really great work.. I don’t care what these people are complaining about.. lets see them print a record HAHA … but great project.. maybe the resolution of the printer will get good enough to have decent audio

    1. Let’s hear your 3D printed record… Oh wait…

      The fact that she’s even able to make it sound that good is a huge feat in itself. I don’t see any of your hacks featured here, so you’re in no place to mock anyone (or their projects) featured on HAD…

  9. Sounds way boomy… RIAA EQ curve would fix that.

    Also, the ‘weee-wooo-weee-wooo’ seems to be aligned with the grain in the model–that’s neat.

    1. It’s because of that curve that this sounds good at all. It is boosting the bass and cutting down all the crap of the poor level of bits at the top end. There should be brick wall filter at about 5 kHz. Good work.

  10. It would have been nice of they’d targeted and used a 78rpm-capable turntable. Or used any DSP to mitigate the 3d printed quantization noise.

    1. (yes, I know she addressed the rpm point, but … 20 minutes of conspicuously bandlimited 11kHz audio isn’t really anywhere near as nice as 3 minutes of 26kHz audio. And there’s no reason you couldn’t use a full 12″ 78rpm for 8 minutes of audio, other than structurality of the plastic your phonograph is made of)

  11. this brings the question. what if you put the raw digital data on the record…. and used some sort of error checking algorithm to fix any errors that might pop up from the printing? MP3 record anyone?

  12. I’d like to congratulate Amanda on her achievement, as well as her excellent taste in music.

    I kinda dig how these records sound.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s