I’ve finally come around to setting up Git but it was a local repo on my computer rather than turning the remote Gallaudet one into a git repository because I will need access to the parent folder. I solved that problem later by just putting all my work into a new folder in that remote Gallaudet network location.

I’ve looked into face-tracking stuff for the 4th setting that everyone wanted so badly. It sounded like a huge challenge, but I managed to implement two different algorithms (face API and TensorFlow.js) but I wasn’t too sure how to incorporate that into our files as it has a different code structure. I asked my mentor for help, but he was not very good with the codes as well. The main issue was that it was very laggy, so it was practically unusable. Suppose we actually got through that issue, I’m still not sure how to do the code for my envision – which is detecting the end of a spoken sentence, change in a speaker or a huge change in positioning to make the dynamic captioning flow with the face tracking well.

Next part of the coding, I worked on looking at RTT which my mentor said that he was going to do but he hasn’t come around to doing it so I moved on to do it. I studied the code and figured out the life cycle of the text – it gets captured from a client, then the client attempts to send it with encoded messages, other clients will receive it and decode it, then they will print and display it on the video. I tried to modify it at every stage but it seems like it was easier to just do it at capture so it doesn’t send until certain conditions rather than not displaying after receiving until certain conditions. The two conditions I implemented are space and enter presses. In hindsight, I think I can add other punucations as well.

I coded scripts that type sentences for me automatically while mimicking backspaces and pauses that humans tend to make. The idea is that it would be a prerecorded video, but for text. This way, the way subjects will always perceive my communication the same down to every pauses and backspace with no variation that could affect the results.

I also made a video montage that quickly showcases our research study’s demos since we have 7 of the conditions. With video editing, I can speed things up and crop out unnecessary parts since my group’s demo study is a lot more involved in terms of setting up than others as it requires an interactive component and 7 servers. Sometimes, I wish I was doing another topic instead of this. No other groups showed their demo sufficiently and detailedly like mine and Dr. Raja particularly liked mine so I guess it was worth my effort. But I saw that other groups just didn’t have many changes from the original i.e. they didn’t create a lot of new things which kind of made me feel bad for doing more work than I had to.

Anyways, I also dynamically captioned a video clip which is the last piece to our pilot study. My partner was having issues with it using Kapwing since it didn’t have crop options, motion for objects, separate transparency, and some other settings. so I used Adobe Premiere Pro to do it.

On the last day of the week, I quickly practiced with my partner for the pilot study we planned with Katja to make sure that we are on the same page and to make sure it goes smoothly. We did it quite well but there are few changes we could make which we will.

One thing they tried to do was a social hour which no one had any ideas what to do. The host asked for recommendations and ideas so I gave two and those were the only thing used – Skribblo and Gartic Phone. Both of them are quite simple drawing games but it wasn’t as fun as playing with friends, probably because that the big guy is here so we feel a little tense. More importantly, we don’t know each other much enough to go like “Haha hey what the heck is this drawing supposed to be? Wait that was a kiwi? No way!” Then we try Gartic Phone which was a bit worse because interpreters joined and the game itself requires everyone to participate but the interpreters didn’t. So there were a few blanks art. Also, it’s a longer game in nature than Skribblo so it was quite awkward for a while. The interpreters also didn’t interpret the occasional comments that I was making so that sucks. All in all, making friends in a virtual environment just hella sucks especially if your boss is watching.