Kokoro Reader

A minimal PDF reader with built-in text-to-speech, powered by Kokoro TTS.

Pure Rust. Runs entirely offline. No Python needed.

Features

• PDF viewing with page navigation
• High-quality neural text-to-speech (Kokoro 82M)
• 50+ voices across multiple languages
• Streaming playback (sentence by sentence)
• Play / Pause / Stop controls
• Voice selection and speed adjustment
• Drag & drop PDF support

Prerequisites

• Rust 1.75+
• espeak-ng (pacman -S espeak-ng / apt install espeak-ng)
• libsonic (pacman -S libsonic)
• pcaudiolib (pacman -S pcaudiolib)
• libpdfium (download from pdfium-binaries)

Pdfium setup

mkdir -p lib/lib
cd lib
wget https://github.com/bblanchon/pdfium-binaries/releases/latest/download/pdfium-linux-x64.tgz
tar xzf pdfium-linux-x64.tgz

Build

cargo build --release

Run

./target/release/kokoro-reader

On first run, Kokoro model files (~340MB) are automatically downloaded to ~/.cache/kokoros/.

Usage

1. Click Open PDF or drag & drop a PDF file
2. Navigate pages with < / >
3. Click Play to read the current page aloud
4. Select a voice from the dropdown
5. Adjust speed with the slider

Architecture

egui (GUI) ── pdfium-render (PDF) ── kokoro_engine (TTS) ── rodio (Audio)

• PDF: pdfium-render for page rendering and text extraction
• TTS: Custom Kokoro ONNX engine with correct Misaki phonemization
• Audio: rodio for streaming playback with play/pause/stop
• GUI: egui/eframe for minimal native UI

Credits

• Kokoro-82M by hexgrad
• kokoro-onnx (Python reference)
• kokoro-tiny (Rust base, patched)
• kokoroxide (Misaki phonemization reference)

License

MIT