catm

On-device long-form text-to-speech in a side panel. Right-click → Read page.

catm reads long-form text out loud, right inside your browser. Synthesis runs locally on your machine — no servers, no accounts, no telemetry. Text you select never leaves the browser. How it works ———————————— 1. Select any text on any page. 2. Right-click → "Read it to me". 3. The side panel opens and starts reading. What's in the side panel ———————————————————————— • Four natural voices, previewable in one click before you commit. • Audio plays as it's generated — start listening within seconds of hitting Generate; the rest streams in as you go. • Text highlights along with the voice, so you can follow or skim. • Playback controls: pause, scrub, 0.75× – 2× speed. • A library of past readings, stored locally. Open old reads instantly. • Pop the panel out into a full tab when you want a wider reading view. How the privacy story holds up —————————————————————————————— • The Kokoro 82M TTS model downloads once into your browser cache (~310 MB) and runs via ONNX Runtime Web (WebGPU when available, WebAssembly fallback). Subsequent reads work fully offline. • Audio is encoded locally to fragmented MP4 and stored in OPFS — the browser's private file system for this extension alone. • No external scripts, no analytics, no remote requests after the first model download. Permissions ——————————— • contextMenus — adds the "Read it to me" entry to the right-click menu. • storage — saves your voice choice and onboarding flag. • sidePanel — opens the reader in Chrome's side panel. Free and open source. MIT licensed. Source: github.com/catm-app/catm-app.github.io Privacy policy: catm-app.github.io/privacy.html