Asymptomatic

Posts tagged: Sn

ActivityPub on Sn

I’ve spent the morning trying to get my ActivityPub implementation off the ground, and I fee like I’m making headway. You can currently follow my account at the blog (@owen@asymptomatic.net), and it seems to record that information for future publishing. The testing suggests that when I publish this post, it’ll push it out to anyone that is following that account, but given that the Mastodon instance that my personal account runs from and my iOS client report different things (Mona thinks that my follow request is “pending”, and I’m pretty sure I’ve set things up for there to be no review of follow requests), I’m wondering what exactly will happen.

Configuration of a system such as this is non-trivial and a bit inscrutable. I’ve tried to narrow options down to the bare minimum necessary, as combined with the other options that are needed for the blog itself, so that it doesn’t take a lot of config to get the system going.

Projects Progress

I have a few technical projects that I’m working on, but they are moving very slowly. Little’s Law has taken effect, where working on multiple things at once makes everything progress at a glacial pace. Let’s talk about the things I’m doing and maybe write out some progress. I might also cover some other topics, like how I’m using current AI techniques to get better at leveraging technology.

One project is called Team Beat. It’s software that organizes a retrospective into phases: a discovery phase where people can submit cards to a board, a voting phase where participants vote on the most important cards, and a discussion phase to talk about those cards. The software takes all the notes from the retrospective and puts them into a markdown document that you can save. The tool works well, and a couple of teams at work are using it. However, there are issues with it. The real-time board submissions are powered by WebSockets written in Go, which connects to the database storing all this information. One problem is that although Go performs well, configuration exists in the front end that really should be part of the back end. This disconnection of config makes it hard to filter down to specific user details in some situations. For example, every action on a board is broadcast on a single channel delivered to all users, requiring them all to fetch new information from the API instead of receiving simple updates.

Post Creation Workflow

While writing this blog, I would like to be able to create a workflow that allows me to quickly create posts written in my own voice and post them to the site without a lot of effort. One of the things that causes a lot of effort to take place is having to type out every single word. These days, it’s pretty easy to use a microphone to record new spoken content and have it transcribed into something that can be posted online. I’ve recently been experimenting with a tool called AudioPen, which records my voice, transcribes it, and then submits it to an AI model to rewrite the content a bit. This is particularly interesting because if you’ve ever recorded yourself speaking live and listened back to it, you’ll note that you make a lot of mistakes, the grammar isn’t necessarily correct, and what you generally hear is pretty forgivable if you’re a listener, but not very forgivable if you are a reader of that transcribed text. I really wouldn’t want to publish directly transcribed text to the website without a significant amount of editing, and so it’s nice that it sends it through this AI model, which corrects some of the grammar and enhances some of the wording to really get at the meaning. Unfortunately, the prompt that AudioPen uses is a little more aggressive than I would like and removes a lot of what makes my voice hear, at least my written voice, sound like me. I tend to sound more like an AI model, so I’ve been playing with some other tools to hopefully make this as easy of a workflow, but with less loss of fidelity in the transcription.

I’ve been exploring a workflow that enables me to quickly generate posts in my own voice with minimal effort. One of the biggest challenges is the need to type out every single word. Nowadays, it’s fairly straightforward to record spoken content with a microphone and have it transcribed for online posting. I’ve been experimenting with a tool called AudioPen, which records my voice, transcribes the content, and then enhances it through an AI model. This approach is particularly fascinating because, as many of us might have noticed, live speech often includes numerous mistakes. The grammar might not be on point, and while these errors are generally forgivable to listeners, they are less so to readers of the transcribed text.

Blog Software Update Reflections

Last night, I modified the blog software running this website. Now, it can load different streams of posts into separate loops on the same page display, allowing a chronological listing in the main section and a random listing of posts in the sidebar that changes with every page load. Impressively, this doesn’t affect site performance – a key goal of mine is maintaining the speed of a static site with static HTML files.

While reviewing the site and ensuring it functioned as intended, I read through some older articles and noticed they generally fall into three categories. The first category dates back to before Twitter, featuring one or two sentence posts with links to other sites – many of which are now dead or defunct, rendering these posts mostly irrelevant.