My Final Project

Auto Record Player

Week 1

I’m a vinyl collector and I enjoy listening to vinyl records, but a problem I face is that if I ever want to listen to an exact song, I have to pick the needle up, find the right groove, and place it there. What if I could automate this process? This would be very useful, especially for compilation albums which don’t have any flow of songs between them. A device that would scan the entire vinyl for deep grooves indicating the various songs. A camera on the tip of the needle will be used to scan the vinyl, and then a mechanised grabber clamper will clamp around the vinyl, spinning it automatically. The needle will then scan the other side of the vinyl as well. This will give the user a full list of songs on the vinyl which will be displayed on a screen, and then the user can choose any song on the vinyl. The gimmick of the vinyl auto-rotating and the needle searching for songs will be part of the device’s allure.

Sketch

Week 2

Designed a concept model for the auto vinyl song chooser. The plan is to make the structure out of dark wood and keep the accents in silver like the buttons, needle arm, and the auto rotator so the whole thing keeps that classic turntable aesthetic.

Fusion Render

Week 4

Set up the ESP32-S3 Sense board with the camera module and managed to stream video to a local server. One key takeaway was that the default camera on the Sense board is the plug-in OV2640 sensor which has a resolution of 1600 × 1200. With this camera it was difficult to clearly pick up the grooves on the vinyl record. Because of that I started considering switching to the OV5640 camera which has a higher 5 MP resolution of 2592 × 1944. With the extra resolution and some diffused lighting it might be possible to capture clearer images of the vinyl grooves.

ESP32-S3 Sense with the OV2640

OV5640

Week 6

Also thought about alternative ways to figure out the exact groove positions in case the camera approach does not work. One idea was to connect to the Discogs API and pull the vinyl details including the length of each song. The issue with that approach is that groove spacing on vinyl records is not consistent. It depends on the specific pressing and the production quality. Some records pack grooves closer together while others space them further apart. Because of that variation the time duration of each song would not directly translate to an exact physical groove position on the disc. Because of this, a direct image processing approach might actually be the simpler and more reliable solution. The ESP32 could capture the images and send them to a local computer where the groove detection and image processing would be done.