Skip to main content


I wish to begin this by saying that nothing can replace the wonderful consideration of people adding alt text to their images. It is very much appreciated, and I thank you for doing it. But for those images without said descriptions, I found something truly useful. It's called AltBot, and it adds alt text to posts without descriptions. I just followed it a few hours ago. On TweeseCake, at least, you'll hear something like "image with no description", and when you go to the next post, you'll hear the bot's explanation of the image. If you wish to folow Altbot, it can be found at altbot@fuzzies.wtf. I don't know who created it, but I sincerely thank whoever did. If only such a lovely bot could work on all sites.

#accessibility #altbot #alttext #blind #blindness #descriptions #Fediverse #Friendica #helpful #images #kindness #tools #TweeseCake

in reply to Georgiana Brummell

Though I'm sure you know of these already, you do say "If only such a lovely bot could work on all sites.", which leads me to think it may be worth linking to the describers for all screen readers. Jaws' picture smart is available in 2024 and 2025, I believe, and basic, though possibly out-dated, instructions are at
toptechtidbits.com/239.html I say the instructions may be out of date only because they suggest that it is still part of the early adopter program. If you don't want to use the Jaws feature, github.com/SigmaNight/basilisk… or, for NVDA, github.com/aaclause/nvda-OpenA… both work well though cost a few pennies for each description. Alternatively, Be My Eyes offers free descriptions from one LLM model.
in reply to Tech Singer

@Tech Singer I use NVDA with the Cloud Vision Ad-on. It's good, but it's nowhere near as good or descriptive as Altbot. There was another one (I forget its name) that used gemini and that was a bit more customisable, allowing a user to write a prompt, set the character limit, etc. , but google updated the version, so the ad-on no longer works. I believe it's available for other sites, though those do require payment.
in reply to Georgiana Brummell

I could never get really good descriptions with CloudVision, I'm not sure what LLM it's using, and gave up on it when I got the OpenAI add-on I mentioned, which has a much wider choice of models, can handle text prompts, and does more than describing images.