book2voice

Turn your EPUB/TXT books into spoken audio locally with Kokoro TTS (WebGPU/WASM). Multilingual.

book2voice turns your own books and text files into spoken audio — locally, in your browser. Load an EPUB, MD or TXT (or paste text), pick a language, and it synthesizes natural-sounding speech chapter by chapter, saving a WAV for each. **Features** - 🎧 High-quality neural speech (Kokoro TTS), running in your browser via WebGPU with a WASM fallback — no cloud synthesis. - 🌍 Multilingual: English, Spanish, French, Italian, Portuguese (BR), and Hindi, with automatic language detection and a default voice per language. - 📚 EPUB and TXT support, split into chapters automatically — every word is included. - 💾 Download audio per chapter with clean, human-readable filenames. - ⏯️ Long books are resumable: close the panel or the browser and pick up where you left off. - 🔒 Private by design: no accounts, no tracking, no analytics. Audio never leaves your device. Only the text needed for pronunciation is sent to a stateless service that stores and logs nothing (see privacy policy). **How it works** Speech is synthesized on your device. Pronunciation (text → phonemes) is computed by a small stateless server, then the audio is generated and stored locally. Model files are downloaded once from Hugging Face and cached. **Note:** Russian is not supported (the underlying model does not support it).