Hello!
I’m a legally blind software developer. I’ve been working in the industry for about 8 years now. While I learned to code the “old school” way, even with my visual impairment, I recently suffered some further vision loss, and it’s exhausting for me to stare at raw code for eight hours a day. So before I introduce what I built, I just wanted to say a quick thank-you to Cursor and its community for making coding fun and accessible for me again. I career changed from academia into software development because I wanted to create products that had a meaningful impact, and Cursor has made that vision (no pun intended) possible. So thank you!
As a legally blind lifelong learner, Wikipedia has always been a bit of a pain for me to access. Sure, it isn’t THAT bad, it’s mostly text, and so it’s possible to browse to the site, search for a topic, and then once in an article engage the screen reader or other TTS. There are also tools like ElevenReader or Speechify to do this, but you still need to know what article you’re loading, etc. Both approaches are doable, but have a little friction.
So I created Curio Garden
Site: https://curiogarden.org
Code on Github: sethwilsonUS / world-garden (the original name before I realized the domain was hella expensive)
Features–largely powered by Wikipedia’s free API
- Users visiting the site will immediately have focus land on the search form. Search for any topic.
- See a list of results from Wikipedia. Arrow through the results and press enter on the one you want to learn about–often the first one
- The article page lets you play audio for the entire article or by section. There’s also an image gallery, good for sighted and partially sighted users. You can also download the article’s audio for offline listening.
- Wikipedia’s featured article is showcased on the homepage, as are their trending articles.
- Get a random article–I have learned so much useless crap thanks to this feature!
- Things like playback speed and listening history are tracked through local storage.
Limitations:
- no auth setup yet, have big plans for dashboards and streaks and such once it is.
- voice is provided by Microsoft Edge TTS. I initially tried to go with ElevenLabs but realized that would be prohibitively expensive, even with their cheapest voice.
- Some sections aren’t suitable for audio, like long tables and such. Maybe have the idea of using AI to summarize these sections.
This is an open-source project, and any comments, constructive criticism, or contributions would be most welcome.
Happy listening and learning!