útoolek

A CLI that turns long audio recordings into clean, readable prose markdown. Uses Qwen3-ASR served through vLLM, with optional word-level timestamps via the matching forced aligner.

Can also use pyannote diarization - to split text by speaker with --diarize flag.

Built around a single-GPU, in my case 24 GB workstation with RTX 3090. On that hardware it transcribes at roughly 150–250× realtime — a 51-minute recording finishes in about 16 seconds after the model is loaded.

Some sites can be addictive. Blocking them completely can be a solution, but sometimes I still want to visit them, just I don’t want to do it mindlessly. This is a userscript that adds 15-second timer before accessing distracting websites.

The timer resets if you switch tabs, so you must wait in foreground.

Inspired by Mindful Internet Use which doesn’t work in Chrome anymore. And Tampermonkey (or similar userscript extension) can be used on mobile: with Safari on iOS and Firefox on Android.

I’m using Niri as my Wayland compositor, and Kanshi to manage monitor layouts. However, editing Kanshi’s config files by handis annoying (though I like to practice mental math a bit).

I wanted a more visual way to manage my monitor profiles. A GUI application for editing Kanshi monitor layout profiles with live preview via Niri IPC.

Drag monitors on an canvas, edit properties, save to your kanshi config, and preview changes live on your displays — all from a single binary.

It’s agentically engineered in Go with Wails (I like it!) and works fine for me, but for sure not all corner cases are covered. If you happen to use Niri with Kanshi, decided to try it out - ping me on GitHub if you have any issues or suggestions.

I just want to keep some structure of the repositories I check out locally. So it just does this. Clones GitHub repos to worspace/[owner]/[repo] and other sources (i.e. GitLab) to prefixed with service name. And then changes directory to the cloned repo. It is a simple wrapper around git clone with some path management.