This semester was not normal. Due to the COVID-19 pandemic, all courses at TU Dortmund University have been conducted virtually. I was part of the team responsible for Grundbegriffe der theoretischen Informatik (basic concepts of theoretical computer science). This course is mandatory for all computer science students and therefore, we had over 850 participants (of which more than 540 handed in the first weekly assignment).
I was part of the group responsible for the exercises. We created and rated the weekly assignments and offered help desks, which allowed students to ask questions regarding the assignments and the course in general. When it came to grading and help desks, we were supported by student assistants. However, in addition to the help desks, we also offered videos explaining possible solutions for exercises in detail.
Creating such videos takes way more time than I expected. Recording the voice-over is not nearly as easy as I expected. I did not count how often I had to repeat some sentences until they were understandable, without long pauses, and grammatically sound. Note that the videos are in German, my mother tongue. Nevertheless, when listening to your own voice many things sound awkward at first. In this post, I want to give an overview about the tools I used and what I have identified to be the fastest way to produce those videos. Keep in mind that everything is self-taught and your mileage may vary. Also, this is not a tutorial on how to use the software, as there are very good tutorials out there.
The first thing I looked for was an easy to used video editor. There exists a wide variety of different (open- and closed-source) video editors that have Linux support. In the end, I settled for kdenlive as it is easy to use and runs well on my (4.5 years) old laptop (thanks to proxy clips).
Since I was not able to get the voice recording working in kdenlive, I used Audacity to record my voice. Here, I found out that the microphone contained in my laptop does not sound very good. Therefore, I bought an external microphone, which really helped to get a nice consistent recording.
To record my screen I used Open Broadcaster Software (OBS), which I already had installed since I streamed my wedding (which also happened during the pandemic and this way all our family and friends could attend—at least virtually). Most of the time I recorded a virtual whiteboard. To this end, I used Xournal++ and a graphic tablet. (If you are using Linux, I recommend Wacom tablets, as they work reasonably well out of the box. I did not get the XP-Pen tablet to work and had to return it.)
As I mentioned before, I will not go into much detail on how to set up all this software, as there are many good tutorials out there, which are also updated when a newer version is release.
For my first video, I wrote down every word I wanted to say. Sentence by sentence. Then I recorded the voice-over. This was actually quite fast to do, because there was a script. However, writing everything down took some time. After the voice recording was done, I created the video. I placed pictures where I wanted them to be and created screen recordings to accompany my voice-over and the pictures.
Creating the first video doing so took quite some time. This is mostly because I was using kdenlive for the first time and using a new (complex) software takes some time.
I continued with this method for my second video. It only took half the time the first video took. Mostly due to my improved knowledge regarding the software. But also due to less unnecessary effects ;-).
After creating the first two videos, I felt reasonably confident in my ability to create the voice-over and tried to record it without any script. This was harder than I thought and I had to repeat some sentences multiple times before they sounded right. Mostly, I had to get rid of some awkward mid-sentence break offs. After I had the voice-over, I created the video to match it. Creating the video was the same as with a script. This resulted in the following video.
After I had created my first three videos, I had four weeks until I had to create the next batch. I had no access to my recording setup for a few days. Therefore, I started with the creation of the videos. Fortunately, this worked really well. I had to cut the voice-over a lot more than before and I also had to change the speed of the recording quite a lot to match the voice-over, but overall this process is not only the fastest but also results in better videos. I think this is because I could create the videos without the limitation of already having a voice-over (or a script) and had a script in form of the video to record the voice-over. Hence, this is overall the best way to create videos (at least for me).