Sonovis kestrel · sv04 Launch Sonovis
The voice made visible

Your PDFs, read aloud — and a great deal more.

Sonovis reads your PDFs and text files aloud with neural voices synthesized on your device. Then it keeps going: word-by-word highlighting, five offline study tools, a sheet-music reader with pedal support, live group listening, and one-tap MP3 export — no account, no cloud.

  • Neural + system voices
  • Five study tools
  • Live group reading
  • Works offline
Read-along Live preview

Word-exact highlighting. In the app, tap any word to start reading from there.

0:00
Your shelf

A library that glows.

Every PDF or text file you add becomes a spine on a shelf you can organize — and every book remembers exactly where you stopped.

  • Spines, not filenames. Hold and drag to rearrange across multiple named shelves; rename and recolor any book.
  • Always resumes. A progress gauge fills as you go, and each book reopens at the exact sentence you left.
  • On your device. Stored locally in your browser — export the whole library to one file and carry it anywhere.
More than a play button

The parts you won't find elsewhere.

A reader, a study companion, a music stand, and a shared listening room — and all of it runs on your device.

Sheet-music reader

Hand Sonovis a score — or any PDF with no text layer — and it opens a visual reader instead of reading gibberish. Sharp pinch-zoom, tap-zones, auto-scroll, and a full-screen fit. Best part: Bluetooth page-turner pedals pair as a keyboard, so you turn pages hands-free while you play.

Pinch-zoom to 4×Pedal / hands-freeAuto-scroll

Read together, live

Start a session and read a book with friends in real time — everyone follows the same sentence and hears it in their own chosen voice. Sessions are private, peer-to-peer, and vanish when you leave. The host shares a code and the book travels with it; a manual code works even with no broker.

Peer-to-peerShare a codeOffline fallback

Export a real audiobook

Render a whole book — or a single chapter — to an MP3 in your current neural voice and speed, with chapter markers written to a sidecar file. Everything is synthesized on your device; nothing is uploaded. Save it, sideload it, listen anywhere.

MP3 outputChapter markersOn-device

Five offline study tools

Recap the story so far, get an Overview of the topics that come up most and the questions the text answers, Search the book, pull standout Quotes, and see the People in it. Everything reads only up to where you are — spoiler-free by default — and every result taps to jump you there.

Recap · OverviewSearch · QuotesPeople
Everything it does

The full toolkit.

A serious reader underneath the headline features.

Karaoke word tracking

Each word lights up as it's spoken, inside the current sentence. Tap any word to start there, and nudge the highlight ±400 ms if it drifts.

Two summarizers

Condense what you've read with an offline extractive summarizer — 5 to 40 sentences — or an on-device AI summary. Read it aloud or copy it.

A shelf of book spines

Uploads become spines across multiple shelves you name. Hold and drag to rearrange, rename, recolor, and watch each progress gauge fill.

Smart PDF cleanup

Column-aware extraction reads two-column pages in order and strips repeating headers, footers, and page numbers — verified against adversarial PDFs.

Playback, out of the way

Sentence and page transport, speed 0.5× to 3× with no re-render, keep-screen-awake, and lock-screen and headphone controls. Every book resumes where you left it.

Three themes, any color

Read in DigiScape, Library, or Galaxy, with a living background you can switch off to save battery. A colour wheel sets any base and accent — spines follow.

Nothing leaves your device

Speech is synthesized locally — no cloud text-to-speech, no account, no keys. After a one-time voice download the whole app works offline; it's a static site you can host yourself.

Backup & restore

Export your whole library — every book, shelf, and exact position — to a single file and import it on any device. Neural voices simply re-download.

Voices

Two engines, your call.

Pick instant device voices, or download a neural voice that runs entirely offline. Sonovis even grades the voices installed on your device and shows you how to get better ones.

Neural AI

Piper models run through ONNX Runtime in your browser. A one-time download (≈63 MB medium / ≈115 MB high, with a live percentage) stores the voice on-device, then it works with no connection.

RyanRyan HQJoeHFCAlanNorthern English

System

Use the voices already on your device — instant, always offline, no download. Word highlighting is exact wherever your browser reports boundaries, with a self-calibrating fallback where it doesn't.

Getting started

Reading in three steps.

Add a book

Load PDFs or paste text

Choose text-based PDF or .txt files — multiple load in order — or paste text straight in. Each becomes a book on your shelf.

Pick a voice

System now, or neural offline

Start instantly with a device voice, or switch to Neural AI and download one once. From then on it reads with no connection.

Listen

Play, study, or share

Follow along word by word, open the study tools, export an MP3, or start a session and read together — settings persist across sessions.

Ready when you are

Put a voice to the page.

Open Sonovis, add a book, and pick a voice. Everything above runs right in your browser, on your device.