This update is about honesty: when you tap a word, you should get a real meaning — not a confident guess. MEANINGS YOU CAN TRUST • Formal written adjectives like 確固たる, 堂々たる and 断固たる showed a perfect reading and then "No entry found". Around 550 words of this kind now define properly. • Words with several genuinely different meanings no longer state the first one as fact. 下げている now reads "is hanging / lowering" instead of only "is hanging". • Newspaper spellings written half in kanji and half in kana — ちょう報, ゆう合 — now find their real entry. • ぴょこっと now has a definition, and is labelled an adverb rather than a noun. SEARCHING IN ENGLISH • Everyday loanwords are findable in English again. "coffee", "salad", "beer" and "restaurant" now reach コーヒー, サラダ, ビール and レストラン — over 5,000 entries restored to English search. GRAMMAR THAT SHOWS ITS WORKING • Negative conditionals used to hide the negative. 来なければ was labelled simply "Conditional" and glossed "if (one) comes" — the opposite of what it means. It now shows the full path, 来る → 来ない → 来なければ, "if (one) doesn't come". JAPANESE NAMES AND EVERYDAY KANA • 丁薛祥, who appears throughout China news coverage, no longer splits into a wrong surname plus a blank. • はい, ここ and バイト no longer come back as "lung", "individual" and "secondary capital city". CHINESE • Chinese words no longer all label themselves "verb", and a Chinese name reads as a name. • Around 7,300 Chinese compounds that showed an empty card now answer. PODCASTS • Long NHK episodes that refused to download — and blamed your connection — now download reliably. Thank you to everyone who reported a wrong meaning. Nearly every fix in this release came from a real report.

Bigger screens, and better Chinese. iPad and Mac • Home and Read now use the width of the screen instead of a narrow strip down the middle. Podcast shelves show more of what they've already downloaded. • Podcast and video transcripts hold a comfortable line length instead of stretching the full window. • Tapping a word in one of your own video files with the device turned sideways no longer cuts the meaning off the bottom of the screen. Traditional Chinese • Sentence patterns like 雖然…但是 and 因為…所以 now explain themselves. They never appeared for Traditional readers before — silently, for as long as the feature has existed. • Tapping a word no longer opens a page spelling it in Simplified. If you have chosen a script in Settings, that still wins. Everywhere • Japanese pronunciation guides stay legible at the smallest text sizes. • Settings is now honest that the data export is a copy to keep elsewhere, not a restore point. On iPhone, everything else is deliberately unchanged.

Toku now works properly on iPad. Podcasts and videos put the player beside the transcript in landscape instead of squeezing it into a few lines — about four times more text on screen. Word cards no longer get cut off at the bottom of the screen. The browser's address bar no longer hides underneath the podcast player. Reading columns, search fields and cards stay a comfortable width instead of stretching across a 13-inch screen. Your place in every book is exactly where you left it.

Toku Reader is now a real iPad app — and read-aloud finally sounds like a person. IPAD • The whole app — listening, reading, the dictionary, your words — now uses the full iPad screen instead of a stretched-up phone layout. BETTER VOICES • Read-aloud now picks the best Japanese or Chinese voice your device has, instead of the basic robotic one. You can choose a different voice for each language, and hear it before you pick, in Settings. LISTENING • Podcast transcripts no longer skip over parts of the audio. Occasionally a stretch of speech was left out and marked as silence; that's fixed, and episodes that were already affected repair themselves. • Shows that publish their own transcripts now come first when you browse, so more episodes open instantly and read perfectly. • Speech-to-text gets names, places and recurring show vocabulary right far more often — and now also recognises words you've already saved. • When an episode genuinely can't be transcribed, Toku tells you why instead of leaving you watching a spinner. TAPPING WORDS • Chinese compound words no longer dead-end at "No entry found". • Pinyin now appears for rare characters that previously showed none. • The word card no longer gives two different answers for words that sound alike — when it truly can't tell, it asks "Which meaning?" instead of guessing. • If you correct a word yourself, the card shows what you changed and what Toku originally said, with one-tap undo. RELIABILITY • A large reliability pass: deleting words, running low on storage, sharing text into Toku, and updating the app are all far safer, and screens now tell you honestly when something failed instead of failing quietly.

Faster, more reliable listening • Trending in Japan, Taiwan, and China — a new row on Home shows what native listeners are actually playing. Nothing to set up. • Pick a show during setup and it now opens straight into a tappable transcript, usually within about a minute. • Many Japanese shows now load a ready-made transcript instantly instead of waiting for on-device transcription. • Episodes that previously refused to play — including many long Japanese and Chinese shows — now play. • The highlighted word follows the speaker far more closely, and tapping a timestamp lands exactly where you expect. • Chinese transcripts now follow each show's own script, so Taiwanese shows read in Traditional characters. • Show, host, and place names come out of transcription spelled correctly far more often. A widget, and one quiet reminder • A new home-screen and lock-screen widget resurfaces a word you're about to forget, along with the sentence you first met it in. • An optional once-a-day reminder when words are ready to review — one calm notification, never a streak nag, and only if you allow it. Reading that looks like a book • Proper Japanese and Chinese typesetting with correct line-breaking, newly bundled Japanese and Chinese typefaces, and chapters that look like chapters. • Paste a Wikipedia or news link and get the clean article. Messy pasted text is tidied up automatically, with one-tap Undo. Your words, reorganized • A cleaner Words tab puts what's due today above your growing collection. • Review cards show the word by itself, then reveal the reading, pitch accent, audio, and the sentence you first met it in. • Every saved word remembers where it came from, so you can jump back to the text, page, or podcast moment. Also in this version: a new look for Damo, our daruma mascot, on the app icon and launch screen; working dictionary lookups inside the reader; clearer word cards for long words; and a fix for a crash when reopening recently removed items.

A cleaner, smarter podcast player. The redesigned player finally shows the full episode title — tap it for chapters and show notes — and adds learner speed presets (0.7× to 1.25×), comfortable always-tappable controls, AirPlay right in the player, and an Undo when you skip a track by accident. Start reading the transcript and the player slims down to a strip so the words get the screen, especially in landscape. Shadowing practice goes deeper: after each sentence the player now leaves a silence for you to speak, then plays it again, a "Shadow this" button sits on the current line, and finishing a drill drops you back into the episode right where you left off. Scrubbing shows the sentence you'll land on, not just a time, and chapter starts gently snap into place. Plus fixes: common grammar expressions no longer show an empty meaning when tapped, Japanese transcripts read run-on speech correctly, and the player's options menu scrolls reliably.

Toku is now podcast-first. The app opens on Home — your shows, front and center. Setup starts by picking a podcast you actually want to hear, and your first transcript teaches the one gesture that matters: tap any word while the audio keeps playing. CREATOR PAGES Tap a show and you land on the podcaster's own page, built entirely from their feed: their artwork colors the page, their description and links are live, their hosts, their trailer, and their support link are all there. Their links — and the links in their episode notes — open right inside Toku, so a creator's whole world is tappable too. ALSO NEW • Chapters appear on the timeline — jump with one tap • Drag to reorder your Up Next queue • Share an episode or a show with a friend • Pull down on a show to check for new episodes • Read a show's full description FASTER AND STEADIER • Much quicker episode opening, and smoother transcripts on long shows • Fixed a freeze that could hit when opening a downloaded episode • Fixed duplicated lines in automatic transcripts • Better recovery when a speech-model download stalls, and changing a show's language now retries transcription automatically

A big update for listening. You can now watch video podcasts with the same tap-any-word transcript as the rest of Toku — and they keep playing as audio when your screen is locked. When an episode ends, the next one you're learning from starts on its own, hands-free, and you can skip forward or back from your lock screen, AirPods, and CarPlay. The Listen tab opens to a curated "start here" library of leveled Japanese and Chinese podcasts (including Traditional Chinese), and you can now loop a single sentence in shadowing practice and tap a show's artwork to jump to its other episodes. Plus fixes: saving a word is more reliable, the podcast directory scrolls smoothly, and more feeds load correctly.

Our biggest update yet. Saved words now review on a correct, proven schedule — words you know come back less often — with a redesigned session: interval previews, one-tap undo, a missed-words re-drill, and a progress heatmap. The reader is rebuilt with generated covers, reading themes (sepia, warm dark, OLED night), and a new "Aa" panel for furigana, pinyin, size, and spacing. New for listening: podcast and video transcripts light up word-by-word as the speaker talks, and you can loop or slow down any line to practice. The web browser gains a curated Explore start page of free Japanese and Chinese sites, clearer menus, and automatic crash recovery. Reading accuracy improved again across numbers, counters, dates, and names — plus stability and polish across the reader, dictionary, and listening tools.

This update makes listening and watching a first-class way to read. • Podcasts are now a full reading surface. Follow shows, download episodes for offline listening, and read a tappable, word-by-word transcript as the audio plays. Practice shadowing by repeating each line. • Watch your own video files with interactive, tap-to-define transcripts — read along line by line at your own pace. • Smoother, more reliable playback. Sentences advance cleanly, no longer replay their opening moment, and the listening controls are simpler — one clear "Shadowing" button. • Review cards now play example-sentence audio, so you hear each saved word in a real sentence. • Faster performance and a range of reliability fixes across reading, watching, and listening. As always, tap any word for its meaning, readings, stroke order, and radicals — and everything core runs on-device, offline.

The biggest Toku update yet: true offline listening, a redesigned word popup, and reading corrections that stick. PODCASTS — NOW TRULY OFFLINE • Download episodes individually, or grab the next 5–10 for a flight or commute • Downloads keep their word-tappable transcripts — read, tap, and shadow with no connection at all • New Downloads manager: every episode, its size, and its reading state at a glance • Browse complete back catalogs with "Load all episodes," now with sharper artwork • Paste feed links from any major podcast host, and more shows load reliably A SMARTER WORD POPUP • Redesigned card with a pitch-accent melody for Japanese and cleaner, expandable detail sections • See alternative readings ranked with their meanings — including name readings — and switch with one tap • "Always use this reading" turns your choice into a personal fix that applies across every text, transcript, and web page — manage them in Settings → My Fixes • Know a reading no dictionary has? Type it in when reporting — it applies immediately READING & POLISH • Reader controls now always appear when you open a text; tap the bottom progress line to bring them back anytime • More accurate Japanese readings for compound words • VoiceOver now reads recognized text in the photo reader, and animations fully respect Reduce Motion • Faster lists and dozens of reliability fixes across reading, podcasts, and the browser No accounts. No streaks. Just reading.

• Sharper Japanese and Chinese reading accuracy. • A more polished, responsive feel — subtle haptic feedback when you save words, bookmark, and turn pages, plus smoother loading throughout. • Improved accessibility: animations now respect your "Reduce Motion" setting. • Clearer feedback when something goes wrong, with a one-tap, fully anonymous way to send a crash report so we can fix bugs faster. • A slimmer, faster download.

Version 5.0.2 is a stability update to 5.0.1, fixing a few issues some readers ran into: • Fixed a freeze that could happen when rotating a video or podcast to landscape. • More dependable importing — EPUB, PDF, subtitle, and dictionary files now handle a much wider range of formats and text encodings. • Sturdier text handling, so words no longer occasionally go missing on unusual or copied-in text. • Faster analysis of long passages. • A range of smaller reading, watching, and listening refinements. Thank you for reading with Toku. If you run into anything, reach us from the in-app support link.

Now you can listen, too. PODCASTS — a whole new way to study • Search and follow 100+ Japanese and Chinese shows, download episodes, and read along with a tappable transcript — tap any word for its meaning, line by line. • Shadowing practice: focus one sentence, auto-pause, replay, and slow it down to speak along. • Listening comforts: background play, a persistent mini-player, sleep timer, skip silence, chapters, and a downloads manager. • Your Library: follow shows, line up what's next, and keep a unified history across video and podcasts. READING IS FREE Reading, watching, and listening are now free and unmetered. On the free plan you can tap up to 50 words a day for their meaning — upgrade anytime for unlimited. ALSO NEW • A faster start: a redesigned welcome, and a one-tap "Read from Clipboard" that reads anything you've copied instantly (nothing is saved unless you choose to keep it). • Curated starter sites on the Browse tab, with a Japanese/Chinese filter. • Pick a translation language for video subtitles, drawn from the video's own captions. • Traditional Chinese improvements, plus many stability and reliability fixes. Thanks for reading — and now listening — with Toku.

Toku Reader 5.0 is the biggest update yet — a complete reading companion across text, web, photos, and now video. NEW — Watch tab: search for a video, tap it, and read along with interactive subtitles. Every line gets furigana or pinyin, the current line follows playback, and any word is one tap from its meaning. A real browser: tabs, history, and a clean reading view for articles — with readings annotated right on the page. Faster lookups: tap word after word without closing the card, and see animated stroke order and radical breakdowns right in the popup. Smarter Chinese: when characters can be grouped more than one way, the word card shows the alternatives — tap to switch. Plus a cleaner five-tab layout (Read · Watch · Browse · Words · Dictionary) and many fixes and performance improvements.

• Sign in with Apple — completely optional. Toku stays fully usable without an account, and you can remove your account anytime from Settings. • Animated stroke order — tap any kanji or hanzi, open its character detail, and watch exactly how it's written, stroke by stroke. • Import multi-chapter EPUBs as books, with per-chapter progress so you always pick up where you left off. • A new page scrubber for flipping through long texts in a flash. • A smoother, more reliable Web Reader — tap-to-define now follows the word as you scroll. • Faster, more accurate readings, stability fixes, and safer handling of your data behind the scenes. Happy reading!

Toku Reader 4.5.6 adds stroke order, long-book reading, an optional account, and a more dependable web reader — plus the usual deep pass on reading accuracy. Stroke order: tap into a character to watch how it's written, stroke by stroke, with a moving brush head. Play, pause, step, loop, and adjust speed. Japanese kanji and Chinese hanzi both render with correct, filled strokes. Read long books: import a multi-chapter EPUB and it now arrives as a single book, grouped by chapter. Each chapter opens fast and remembers your place, and your overall progress is tracked across the whole book. Faster navigation in long texts: a new page scrubber lets you drag to jump anywhere in a piece, with a live page indicator. The reader controls now sit clear of the home indicator. Optional account: sign in with Apple to secure your account — it's entirely optional. Everything works without one, and your reading library, saved words, and progress live on your device either way. You can sign out or delete your account at any time, and export a full backup of your data as a file. A more dependable Web Reader: annotation is more reliable across sites and single-page apps, the word popup now follows the word as you scroll instead of disappearing, and tap-to-define keeps working even on pages that load slowly. Reading accuracy: a large round of Japanese and Chinese corrections to readings, word boundaries, and meanings — drawn from real corpora of literature and journalism. Stability and your data: fixed a rare launch issue, and made data updates safer — if anything goes wrong while updating your library, the app now keeps a safety backup automatically instead of starting fresh. Plus many smaller polish items across the dictionary, reader, and settings.

Toku Reader 4.5.5 polishes the popup, deepens the library, and quietly improves how it reads almost every text. Dictionary popup: see every matching dictionary entry, not just the first. Entries are ranked by frequency so the reading you actually want is on top. Multi-word expressions surface ahead of their parts, and the layout has been tightened for legibility. Continue reading, anywhere: the app now remembers where you left off across every reading surface — pasted text, web pages, OCR scans, and library works — and shows a subtle cue when you come back. You can also mark a piece as finished by hand. A larger Japanese library: many new classic short stories and essays have been added. Web Reader: more reliable tap behavior on Japanese and Chinese pages — tapping a character no longer accidentally follows links. OCR: when you save text scanned from a photo, the position you tapped is preserved so you can pick up exactly where you were. PDF imports: the layout engine handles vertical (tategaki) text more reliably, with better column ordering and furigana filtering. Tokenization accuracy: hundreds of small fixes to readings and word boundaries, drawn from real corpora of Japanese and Chinese literature. Library capacity raised: you can now keep up to 100 texts, with up to 100,000 characters per text. Plus dozens of smaller polish items across the dictionary, reader, and settings.

Pick up exactly where you left off. The reader now tracks your reading more carefully — a subtle marker shows where you stopped, and reopening a text returns to the precise token, not just the page. The same continuity works across the paste reader, web reader, and OCR. The reading library has a new home. A Home / Library segmented control replaces the older browser. Home surfaces a Continue Reading rail and recommended works; Library adds a search bar and filter chips for language, collection, author, era, level, and sort — making it easy to find something to read at the right difficulty. Dozens of new short stories, fables, and poems added — early-20th-century Japanese literary works alongside classical Chinese poetry and prose. All bundled offline. Behind the scenes: thousands more dictionary entries covering verb-form variants and uncommon spellings. Sharper word boundaries in pre-war and literary text. Several hundred previously-empty word definitions across the bundled library are now populated. Other improvements: - Pasted text limit raised to 100 saved texts and 100,000 characters per text. - Web reader: taps near images, decorative spans, and mixed-script runs now reliably land on the nearest real word. - OCR: tapping a word and saving the photo to your library carries that tap as the resume anchor. Everything still runs offline. Your reading stays on your device.

Improved reading accuracy across Japanese and Chinese texts — especially pre-war and classical literature, long-form non-fiction, and modern serialized fiction. More words are now segmented and pronounced correctly on the first read, with fewer manual lookups needed. Everything still runs offline. Behavior, permissions, and your saved data are unchanged.

Maintenance + accuracy update. ENGINE READINGS — 29 IMPROVEMENT SESSIONS Japanese (15 fixes): Better tokenization in literary prose, news articles, and Aozora classics. More accurate readings for proper names, conjugation chains, and katakana compound nouns. Specific articles triaged this release: Asahi (磐越道バス事故, ADHD procrastination), NHK (憲法施行79年), Akita Ujaku (三人の百姓), 歳月の門 Edo merchant series, 奥日光・神猿, 狐の嫁入り 第41-84話, てくてく日本語 podcast batch. Chinese (14 fixes): Classical narrative (三國演義 chapters 21-23), more accurate Tang poetry (李白, 杜甫, 白居易, 王維, 孟浩然, 王之渙), and modern news clusters. Polyphone disambiguation for context-sensitive characters. Better sandhi handling within compounds. Specific articles triaged this release: BBC Chinese (FIFA-CCTV broadcast, Iran Lego-AI, AI-brain feature), 轉角國際 (長崎蛋糕之妻), UDN (洋妾花終焉怪談), BBC Anta sportswear, 三國演義 ch.21-23. READ-ALONG AUDIO TTS now skips punctuation tokens during read-along, so spoken pauses sound natural — no more reading 「 」 as literal characters. REGRESSION NET Every engine fix in this release was verified against a curated corpus of 29 articles spanning Tang poetry, Bashō haiku, Showa novels, modern news, Three Kingdoms classical narrative, and Taiwan tabloids — ensuring nothing previously correct got worse. WHAT DID NOT CHANGE - No new permissions, background modes, or third-party SDKs - All reading data still stays on your device — no analytics, no sync - Same OTA delivery, same in-app purchases, same UI Thank you for using Toku Reader. If you spot a wrong reading on a token, please send feedback through Settings — every report makes the engine smarter.

v4.5.1 — Reading quality + content updates • Engine accuracy: dozens of small triage fixes across the Japanese and Chinese readers — better proper-noun handling, cleaner katakana long-vowel readings, smarter polyphone resolution, tighter idiom/counter merges, and improved POS classification for function words. • Pull-to-refresh on the Read tab, plus a "Refresh content library" button in Settings — manually fetch the latest curated works on demand instead of waiting for the daily auto-refresh. • Content library additions: Three Kingdoms (三國演義) chapter 2, 白居易《問劉十九》, and editorial polish across existing works. Thanks for reading.

A curated reading library, all-new in this update. NEW — CONTENT LIBRARY Browse hand-edited classic works with line-by-line English translations, footnotes, and inline cross-references. Tap any character for the same instant dictionary you get on your own pasted text — but with editorial glosses on top. Launching with: • Japanese: Bashō (古池や / 閑さや), Issa (雀の子), Akutagawa (蜘蛛の糸 / 杜子春), Miyazawa (やまなし / 注文の多い料理店), Niimi Nankichi (ごん狐 / 手袋を買いに), Ogawa Mimei (赤いろうそくと人魚) • Chinese: Tang quartet (Li Bai 靜夜思, Wang Wei 鹿柴 / 鳥鳴澗, Du Fu 春望 / 絕句, Wang Zhihuan 登鸛雀樓, Meng Haoran 春曉, Bai Juyi 問劉十九), Han Yu 師說, and the opening chapter of 三國演義 A single "Read" tab now hosts both your imports and the Content library. New works arrive over-the-air — no app update needed. NEW — INLINE EDITORIAL APPARATUS Where the editor flagged a phrase, you'll see a small ※ for a footnote, an EN chip for a translation bubble, or a ↗ arrow that jumps you straight to the work it references. Toggle them off any time if you want a clean read. ENGINE Twenty-plus Mandarin segmentation and tone fixes triaged from real-world news pages — proper-noun pinyin title-casing, 會面 / 面臨 split, country-abbreviation handling, 中 / 以 / 為 polyphones, and 和美 / 三星 / 和數 disambiguation. Japanese engine adds yamanashi-class onomatopoeia (クラムボン, かぷかぷ) and fox-story compound nouns to the user dictionary. POLISH • Reading position is preserved when a Content work updates mid-read • "Seen in Content" row in the dictionary surfaces every place a word appears in the library • Words tab gains a Source filter (Reader / Content / Imported) • Dynamic Type support tightened across reader popups • Several deprecation and stability cleanups ahead of iOS 18 Everything still runs offline — tokenization, dictionaries, OCR, SRS, and now the Content library too.

v4.4 — Sharper readings, smoother browsing. Japanese engine • Native readings for 1人 / 2人 (ひとり / ふたり) and a full counter audit pass • NHK news headline accuracy: 大統領, 錦織圭, 高まる, 年越え and dozens more • Better handling of 行く forms, common compounds, and onomatopoeia definitions • New pipeline-level regression coverage for device-log feedback Chinese engine • New Wikidata-backed proper-noun pack: countries, capitals, Greek and Roman deities, Western historical figures, and Taiwan politicians now segment and define correctly • Smarter Traditional / Simplified mixed-script handling for tokens like 担憂, 脱鉤, 現状 • Optimal-segmentation rewrite eliminates over-merge regressions on long Taiwan literary passages • Numbers and punctuation: cleaner segmentation, English glosses for digits, tone-marked pinyin preserved • Closed out the 為 passive-marker compound family and several non-CEDICT four-character expressions Web Reader • NHK ONE, BBC, NYT and other single-page article sites: furigana and pinyin now persist across in-page navigation. • Page language detection is now per-block: mixed Japanese / English / Chinese pages annotate every block correctly • Reading Aid (Off / Primary / Furigana / Pinyin) survives single-page-app route changes; no more silent reverts to Off Popup • Taps now open directly into the expanded word card — no compact-tier flash, snappier feel • Cross-token taps animate cleanly; the card never covers the tapped word OCR Reader • Language picker now honors your choice end-to-end, not just at the recognizer • Selection persists across sessions • Old-form kanji (旧字体) and IVS variants handled correctly in tokenization Stability and polish • Numerous device-log feedback fixes across both engines • Documentation and architecture refresh for v4.4 No new permissions. No new background modes. Everything still runs offline on your device.

Traditional Chinese: - Shinjitai fold layer improves tokenization and dictionary hits on Traditional-only and mixed-script text. - IVS / surrogate-pair handling fixed (emoji, variation selectors no longer break tap targets). Reading Aid: - New 4-state Reading Aid picker (Off / Primary / Furigana / Pinyin) on every reader + Settings. - Pinyin sandhi-display toggle for learners who want to see the underlying tones. OCR: - Rebuilt on Google ML Kit (on-device). More accurate on photographed print. - Language-override picker if auto-detect picks the wrong script. Web Reader: - Ruby/pinyin/furigana now render reliably across news sites and any page background. - Headlines (H1-H3) are annotated. WaniKani: - Image-only radicals render correctly (one-time re-sync on first launch). - Kana converter, English-shake detection, mined-word tap polish. Chinese engine: - 20 high-confidence polyphone fixes (surname vs. common-word). - Country-pair anchors, modern defaults for 於 / 讀 / 读 / 南.