Local PDF reader turns books into audiobooks on-device
This is an in-progress Tauri 2.0 desktop app for Mac that reads technical PDFs aloud entirely locally, using Kokoro 82M for text-to-speech and llama.cpp/Qwen for optional rewriting or podcast-style exports. The core experience is the tight loop between playback and reading: the current sentence stays highlighted while audio plays, so the app feels like an augmented reader rather than a detached converter. The main work now is making chunking, alignment, and first-run latency good enough for long technical books, especially when the source includes code blocks, tables, and other layout-heavy sections.
Strong niche and a clear user pain point: it is more compelling than generic “PDF to audio” tools because it preserves the reading workflow instead of replacing it.
- –The on-device stack is the right story here: privacy, cost control, and no cloud dependency are real advantages for technical book workflows.
- –Live text highlighting is the key differentiator; without it, this becomes just another TTS pipeline.
- –The hard parts are the right ones: sentence segmentation, PDF text extraction, and keeping speech aligned with the source text.
- –The 15-sentence prefetch idea makes sense, but the real goal is predictive buffering that hides warm-up latency without making the app feel laggy.
- –The podcast-style rewrite mode is a good second export path, but it should stay optional so the core “read my book aloud” use case stays fast and faithful.
DISCOVERED
3h ago
2026-04-29
PUBLISHED
4h ago
2026-04-29
RELEVANCE
AUTHOR
purellmagents