ReadTune

A quick check suggests reading settings to try, then applies them to any article, PDF, or web page. On-device read-aloud.

ReadTune is a free reading tool. A short check suggests a reading setup worth trying — font, spacing, contrast, pacing — and then applies it to any article, PDF, or web page. No account. No subscription. No extension analytics. Your reading profile stays on your device, and read-aloud runs on-device by default. It keeps working offline. ———————————————————— THE PROBLEM Many people find dense digital text tiring or difficult to track, including some readers with dyslexia, ADHD, low vision, or eye strain. ReadTune is built to reduce presentation friction without claiming that one setup works for everyone. Tools that help do exist: roomier spacing, calmer colour, a cleaner layout, text read aloud. Two problems. First, WHICH of those helps is different for every person, and most tools just hand you twenty toggles and let you guess. Second, the tools that do the most put the useful parts behind a subscription or a school-district licence — so a student who needs this and has no money or school login gets nothing. ———————————————————— WHAT READTUNE DOES DIFFERENTLY It opens with a short check — a warm-up plus six short passages, about four minutes, or skip it and start reading on a evidence-informed default. Standard text appears twice on different passages; the other four passages each change exactly one thing versus that default: the font, the spacing, or leading-letter bolding. The order is shuffled. For each one it notes your reading time, a fill-in-the-blank check that a skim can't pass, and a 1–5 "how did that feel" rating. Passages don't repeat if you retake it. Then it compares every change against your own baseline — so a naturally slower reader isn't penalised — allows for the practice speed-up, and keeps only the changes that clear a real margin. If nothing clears the bar, it tells you that instead of inventing a winner. The result is a reading profile: your font, size, spacing, bolding, tint, and pacing. Reader View, PDF mode, and the in-place page restyle all render through that one profile, so what the check suggested is what you get everywhere. ———————————————————— EVERYTHING IN IT — ALL FREE Reading and layout • Fonts chosen for legibility: OpenDyslexic, Atkinson Hyperlegible, Lexend, or your system sans • Text size; line, letter, word and paragraph spacing; line width; a gentle contrast control • Leading-letter ("bionic") bolding, adjustable • Automatic hyphenation, plus real syllable breaks you can show inline (in·for·ma·tion) or pull up one word at a time — double-click any word for its syllables and a "hear it" button • Nine calm reading tints plus a custom colour; an optional line tint that alternates a faint wash between lines to help you keep your place down the page • Remove italics, hide images, freeze animated GIFs Focus and pacing • An adaptive 1 / 3 / 5-line reading ruler that follows your line and dims the rest • Paragraph focus — everything but the paragraph you're on is dimmed • One sentence at a time, stepped with the keyboard or the transport bar • Speed reader — one word at a time, adjustable words-per-minute • Auto-scroll at your reading pace Read aloud • A natural neural voice (Piper) that runs entirely on your device. The default voice ships inside the extension, so it works offline with nothing to download and no permission prompt • The current sentence and word are highlighted as it speaks, in your chosen font • "Voice Fit" in the Reading Lab lets you preview a few on-device voices and keep the clearest; extra voices download a one-time model from Hugging Face and are then cached on your device • Optional "Premium voice" — a hosted voice through ReadTune's relay, no account and no key. Each sentence as it's read is sent to synthesise it, plus the next one prepared a moment ahead; it falls back to the on-device voice automatically if the relay is busy • Optional: bring your own ElevenLabs API key for a different voice. Your key is stored only on your device; the passage is sent only to api.elevenlabs.io while reading. Removing the key also returns that optional host permission Talk to type • Dictate into normal editable text fields on a page — email, docs, comment boxes — with spoken punctuation ("period", "comma", "new line") • Password fields are intentionally blocked • Uses Chrome's own speech recognition, which sends microphone audio to Google to transcribe. ReadTune does not record or store the audio or transcript Reading help (optional AI) • Ask AI is available in Reader View and PDF mode for summaries, article/PDF questions, definitions, explanations, and annotations • Opening the AI rail sends nothing. Where Chrome already has a compatible on-device model ready, ReadTune uses it. Otherwise, only the relevant article text — or relevant text extracted locally from the PDF — and any typed Ask question are sent through ReadTune's AI relay to generate the response • The PDF file itself is never uploaded • Only article summaries are eligible for ReadTune's shared 30-day response cache. Typed Ask answers, Define/Explain results, and annotations are not stored in that cache • Plain-language "Simplify" for a passage you select is on-device only; if the browser has no ready model it says so rather than sending your selection anywhere Where it works • Reader View — pulls the article out of any page and re-renders it • "Restyle this page" — reformats the page you're already on, in place, with a small floating bar; toggle off to restore it exactly • If the page publishes its own "Listen to this article" narration, ReadTune can surface that source audio rather than reading over it • PDF mode — extracts selectable text locally from a worksheet or handout and reads it through the same engine; the file itself is not uploaded Accessibility of the app itself • A "dyslexia-friendly menus" switch makes ReadTune's own buttons, sliders and settings roomier and uses the high-legibility Atkinson/Lexend UI stack Memory • Your profile and history are saved with the browser's local storage — no account • The Reading Lab keeps recent checks so you can see what's stable versus a close call • Reader View remembers where you left off and keeps your highlights, per page ———————————————————— WHAT WE'RE HONEST ABOUT ReadTune is informed by accessibility research. Read-aloud with follow-along and spacing have stronger support than many visual preference features. Dyslexia-specific fonts, coloured overlays and bionic bolding have mixed or weak evidence — some people clearly prefer them, but that's comfort, not a cure. ReadTune labels each feature by how well it's supported and treats the weaker ones as experiments you opt into. The check is a quick preference estimate from six scored short readings plus a warm-up. It is not a diagnosis, clinical assessment, medical device, or treatment, and it does not guarantee faster reading or better comprehension. ———————————————————— PRIVACY No account. No extension analytics or telemetry. Nothing is sold. Your reading profile, check history, reading position and highlights are stored with Chrome's local extension storage, on your device. The extension installs asking for activeTab, scripting, storage, and one required host permission for ReadTune's own relay on readtune.tech, used only by explicitly requested cloud AI actions and the optional Premium voice. Read-aloud uses the bundled on-device Piper voice by default. Optional extra on-device voices download a model file from Hugging Face once; the reading text is not sent for those local voices. Optional features send data only when you choose them. Talk to type uses Chrome's built-in speech recognition, which sends microphone audio to Google to transcribe; password fields are blocked. ElevenLabs read-aloud, if you add your own key, sends the passage and key directly to your ElevenLabs account. Premium voice sends each sentence as it is spoken to ReadTune's text-to-speech relay, which forwards it to a third-party voice provider and does not store the text or generated audio. Ask AI is available in Reader View and PDF mode. Opening the panel sends nothing. Where Chrome has compatible on-device AI already ready, it can run locally. Otherwise, the relevant article text — or relevant text extracted locally from a PDF — and any typed Ask question are sent through ReadTune's relay. Current relay infrastructure/providers may include Vercel, Upstash Redis for summary caching/rate limiting, OpenRouter and its configured model providers, and Ollama Cloud when configured. The PDF file itself is never uploaded. Only article summaries are eligible for the shared 30-day response cache; typed Ask answers, Define/Explain results, and annotations are not stored in that cache. The article URL is not sent to the AI relay. For abuse prevention, the relay derives a short pseudonymous rate-limit identifier from the requesting network address; the raw address is not written into ReadTune's Redis rate-limit key. Free and open source. The full code, Terms, Privacy Policy, and third-party license notices are available from the ReadTune project.