Any document.
Read aloud.
Turn PDFs, e-books, Word files and notes into natural narration, and follow along sentence by sentence. Optionally, create a voice of your own. Everything runs on your computer: no accounts, no cloud, no telemetry.
The 45-second overview. Its voice-over was generated by Local TTS.
Everything you need to listen
A reader, a voice library and an optional voice studio, in one app that you run with two commands.
PDF, EPUB, Word, Markdown and text
Upload a file or paste text. Chapters and pages are found for you, and you pick up where you left off.
Follow every sentence
The sentence being spoken is highlighted, and the page follows the narration from chapter to chapter.
Starts in about a second
Sentences are generated just ahead of you and cached, so even whole books start playing almost at once.
54 natural voices
Kokoro voices in nine languages, with quality grades and side-by-side comparison on your own passage.
A voice of your own
Voice Studio clones a voice from about 20 seconds of speech, with quality checks and required consent.
Stays on your computer
Binds to 127.0.0.1, works offline after setup, and never sends documents or recordings anywhere.
See it working
Recorded from the running app with its real narration: a PDF upload, read-along across chapters, voice comparison, Voice Studio, export and the dark theme.
The cloned voice in the demo comes from a built-in synthetic voice, not a real person. How these videos were made
Get started
You need Node.js 22.22 or newer and Git. The first run installs everything else inside the project folder: a checksum-verified Python toolchain and the voice model (about 340 MB). No GPU is needed.
git clone https://github.com/carsonmcrae/local-tts-web-ui.git
cd local-tts-web-ui
npm install
npm run dev
Your browser opens at http://127.0.0.1:5173. The
user guide covers
installation, voices, voice cloning, exporting and troubleshooting.