// a weekend project · built with Claude Code

Make reading interesting again.

Articles and books, read aloud by a neural voice — entirely on your Mac.

Hear it

A real Summary-mode read — local LLM + CSM-1B, in codeword, a custom-tuned clone voice.

0:00

From a pasted URL to replaying it days later — the terminal does everything.

Open it — paste a URL; /lib, /voice, /model, /mode do the rest.

Every read saved to a searchable library — replay from any browser on your network. Deploy to a Raspberry Pi and it's always on.

Web dashboard — search, sort, and replay past reads.

Home server — Readback on a Pi via PiZoW, always online.

✓offline100% on-device — extraction, summary & speech run on Apple Silicon. No cloud, no API keys.
✓sourcesSnap a book, hear the chapter — read a URL, an image, or a folder of page photos; a local vision model OCRs them into one continuous read, and the tone adapts — a link sounds like an article, a book scan like a book.
✓voiceA voice worth listening to — CSM-1B neural TTS (Sesame); clone any voice from a short clip, or LoRA fine-tune.
✓modesFull or Summary — verbatim, or a local Ollama LLM rewrites it as a spoken explanation; /model fits your RAM.
✓playerA real player — pause, seek ±5 s, and a transcript that highlights word by word — in your shell or the web dashboard.
✓libraryReplay anytime — every read saved to a searchable SQLite library.
✓networkRun it on your network — deploy the dashboard to a Raspberry Pi (built on my PiZoW setup) and replay from any device at home.

Install steps, the pipeline, the architecture, and the full build story — all on GitHub.