One of my favorite new self-hosted apps in 2024 was Hoarder, a "self-hostable bookmark-everything app" which supports archiving and AI tagging features. Over the years, I've used many such apps and have somehow managed to always export and migrate my links each time before they eventually shut down. I'm pretty sure I have bookmarks going all the way back to the del.icio.us days. Most recently, before switching to Hoarder, I was using Omnivore.
Unfortunately, omnivore announced they would be winding down their service and creating a new app specializing in Text-to-Speech (TTS). This is a good idea – in fact, I've been running TTS on all of my saved articles and turning them into a personal podcast feed for many years. I started with the mimic3 TTS engine before switching to StyleTTS. My version was an unholy pile of python and bash, but it did the job!
Over the 2024 holiday break, I greenfielded the gist of those scripts into something I was willing to publish so hopefully others can self-host their own TTS pipeline.
I'm very excited to announce HoarderToPod and its companion FlaskTTS.
The basic workflow is simple: add an article to Hoarder as normal. Then the HoarderToPod service polls for new articles, submits them to FlaskTTS for processing, and finally compiles the MP3s into an RSS podcast feed.
This project is still in beta, but I've been using it for over a month. Currently, FlaskTTS supports StyleTTS2 and KokoroTTS.
If you try out HoarderToPod, please let me know and/or submit any issues to the repository.