Progress report Oct 2023



It's been a hot minute since I last updated the site. But since then quite a bit has happened. First there is the plan9 bootcamp. I learned about a bootcamp that SDF was running during the fall of 2023 from a youtube channel I follow (adventurein9) and decided I may as well join it. It isn't the first SDF plan9 bootcamp I have been aware of, but since then I have decided that it would probably be a good idea to try to involve myself in the community a bit more than none at all. It has had middling success. It hasn't been totally useless, but I could probably try to make a bit more of an effort. Thanks to the catalyst of the bootcamp however there have been some priority changes, since I wanted to make an SDF web landing page, I decided that muffet should have higher priority. Up to the recent past muffet only had the ability to produce gemini pages from the indental formatted database backend. Before muffet, I had been using an old C version of oscean for my static site generation. While I liked a lot about how it worked there were some obvious deficiencies, so I decided a greenfield implementation would be best. Muffet still shares a lot with how oscean did things, but it is 100% written from scratch and only copies the parts I actually mean to copy. It took me a few days to add the html builder function. The bulk of it was done on the first day, but it took a while to work out all the minor issues I had in the html generation, and to get a CSS file to work with it that I was more or less happy with. For a long time I had on the far back burner the idea that I would eventually get around to doing voice recordings of all the written works on the site. There have been a few I did in the past, and a few got videos of them. But to say I haven't been satisfied with the lack of progress on that front would be a bit of an understatement. During some of my youtube perusal I have noticed a marked uptick in the amount of AI generated videos. The voice quality is never "great" but it's clearly better than past TTS options, so I decided to look into how to do it locally. I am not in anyway interested in using some website where you upload the text then download the audio. For me if I can't run the network locally it may as well not exist. I was able to find one that I could run and spent a while learning how it worked and playing with it, then decided to use my album ON 2 as the source to create a voice for reading. The result was satisfactory. It takes a bit of modifying my natural typing in order to make certain things produce the right sounds, but after learning the basics, it's generally pretty good. So at this point I have spent the last week or so feeding the neural network all the various writings on the site and generating wav files, then manually converting the wavs into mp3s. No one ever said neural networks ran super fast. Anyway, as I writing this, all the longer pieces have been vocalized and I only have a smattering of smaller works to complete. Then there is a little bit of book keeping to do with muffet, since I have decided I want to add a footer to the generation with a bit of info to include on every page. One of the boons of static site generation, spend 5 minutes to change over 100 files. There is of course still changing the database files to include the sound file links. It's a simple line like {^snd filename} to include them in both gemini and html. but since we are talking about adding it to every single text on the site, it is no small number. While doing the above I noticed I was a bit lazy when I first added the tagging system to my database, and didn't create taglists for all the texts on the site. Since I want every text to either have the "safe" or "nsfw" tag as is appropriate, at the very least, I need to actually do a lot more housekeeping than I previously expected. By the time this page is up this all should be done. .

incoming references

F 00100 REBOOT