<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Shuush — Engineering &amp; Benchmarks</title><description>Open-source local Superwhisper &amp; Wispr Flow alternative for Linux/Wayland, built on voxtype.</description><link>https://shuush.kitsunekode.in/</link><item><title>Why Raw LLM Dictation Cleanup Fails (and the Span Gate That Fixes It)</title><link>https://shuush.kitsunekode.in/blog/anti-paraphrase-span-gate/</link><guid isPermaLink="true">https://shuush.kitsunekode.in/blog/anti-paraphrase-span-gate/</guid><description>Piping Whisper or Parakeet transcripts straight into an LLM leads to silent rewrites and poisoned dictionaries. Here is the deterministic span gate and phonetic self-learning architecture behind Shuush.</description><pubDate>Sun, 22 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Replacing Superwhisper&apos;s 10k Word Cap on Omarchy &amp; Hyprland with Local GPU Dictation</title><link>https://shuush.kitsunekode.in/blog/omarchy-hyprland-superwhisper-alternative/</link><guid isPermaLink="true">https://shuush.kitsunekode.in/blog/omarchy-hyprland-superwhisper-alternative/</guid><description>A clean, zero-fork guide to running unlimited local GPU dictation on Arch Linux, Omarchy, Hyprland, Sway, or Niri using stock voxtype-bin and Shuush.</description><pubDate>Sun, 22 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Why We Built Shuush: An Open-Source Local Superwhisper &amp; Wispr Flow Alternative for Linux</title><link>https://shuush.kitsunekode.in/blog/why-we-built-shuush/</link><guid isPermaLink="true">https://shuush.kitsunekode.in/blog/why-we-built-shuush/</guid><description>How we built Shuush cleanly on top of MIT voxtype for Linux/Wayland: 19,498 real dictated words benchmarked, 5-source AT-SPI context, anti-paraphrase span gating, and phonetic self-learning.</description><pubDate>Sun, 22 Nov 2026 00:00:00 GMT</pubDate></item></channel></rss>