Hi, person who wants to keep running the objects while I’m gone! This is a good document for you to have. I will try and do my best in documenting all the necessary information, but if in doubt just email me.
Each object contains a raspberry, the PCB with the wires connecting to the button, as well as the microphone. The p2p objects contain a pair of headsets as well.
For my thesis evaluation I deployed four pairs of objects around campus - now only two pairs are around the lab, which I describe below. The rest of the Raspberries can be repurposed, along with the microphones, headphones, PCBs etc. I have left a bunch of prototyping materials (buttons, dials, different microphones etc) on the black wire shelf. Everything there is up for grabs!
The Raspberries all have a number (taped above the Ethernet plug). The idea initially was that Raspberries 1-5 were going to be p2p, and raspberries 6-10 were going to be Twitter objects. Obviously that was too hopeful to be true, so I never stuck to it. Due to some PCBs not working well, two Raspberries were not deployed at all. Here is the spreadsheet with the final characterizations, and contains a bunch of useful comments too.
In terms of code, this is my github repo: https://github.com/Kallirroi/raspi_code
It’ a lot, I know, but all of the Raspberries should have all of that installed. The most important part is the last lines where I describe the pm2 workflow.
The button legs break often (unfortunately that requires changing the button and soldering again)
The button caps also break (but that’s easy to fix, I have 3D printed a bunch of extra ones) People might leave the object on record mode (you can tell if the PCB LED light is on). That will at some point crash the pm2 process, which you will need to restart (
pm2 flush && pm2 restart main —update-env —log-date-format ‘DD-MM HH:mm:ss.SSS )
You don’t really have to worry about the dat part, since we only have a single pair of objects working. Ignore these parts in the spreadsheet as well as the README on Github.
The IP is dynamic and changes, so if you plug a Raspberry back it won’t have the same IP address indicated in the spreadsheet (raspi2 and raspi3 might since they have been continuously plugged in, but I wouldn’t bet on it). You need to connect it to the monitor and look up the
To fix: the button on the p2p object is broken (need to change it).
Given the broken button, I have disabled the recording ability for the moment (on the p2p object). It should only play back audio and not understand when the button has been pressed. If you are fine with that, you can just let it be.
The Twitter object should work as normal - note that the Google Speech to Text is quite sensitive to accents, ambient sound etc and can often not recognize what was said. Persistence is key!
These are not plugged in, so leave them. They should be functional when plugged in though, in case that is ever needed.