ToneWickCHINESE, ILLUMINATED

Chinese, one tone at a time.

Read, listen and explore. Your library stays on this device.

Warming up the reading lamp…
Character size
42px
Pinyin size
22px
Tone size
36px
Website size
130%
Ready when you are.
0.8× gentle · 1× natural

A little light, a little Chinese

Page 1 / 1

Display changes never overwrite your original text.

Tap a word for its meaning. Space to listen.

A Chinese reader that makes pronunciation visible

ToneWick · 聲芯 is a free browser tool for reading simplified and traditional Chinese. See pinyin, zhuyin and dictionary tones, tap words for definitions, and listen to Mandarin. Resize characters and tones independently, or choose gentle starter narration and follow the spoken characters.

Paste your own Chinese or import text and selectable-text PDFs. Your library stays in this browser, with no analytics or advertising trackers. Starter stories and recordings need no account. Preparing new narration requires a separate access key and consent.

A voice for your page

A connected Mandarin reading, prepared on ToneWick’s server. First preparation takes a little time; playback then flows without waiting for tone animations. This is a synthetic voice.

The voice speaks at this pace as the recording is created. Letter timing is matched to the new recording.

Word and character timings are estimated from the audio. Where individual timing is unavailable, a still passage marker follows the recording.

Text to be sent · this reading page only

Generated audio is held in this tab. The server copy is deleted after download where possible and expires automatically. Closing this panel does not cancel a queued job: use Cancel below.

Connect the self-hosted voice to begin.

A WORD, OPENED

Bring your own Chinese

Copy Chinese from a website and paste it below, or choose a PDF or text file. We turn its text into the same rich reader. Your file stays in this browser.

0 / 100,000 UTF-16 units

Plain text files: up to 2 MB. Website formatting and scripts are discarded; only text is read. Nothing is silently truncated.

Switch freely later. The original pasted or extracted text is kept unchanged.

Your unchanged source

This is the original pasted or extracted text, before display-script conversion. Copy it freely. For PDFs, it is the reviewed text layer, not the PDF file itself.

Your small library

Saved on this device, in this browser. Keep a backup before clearing browser data.

Your word pocket

Two paths. One reading desk.

Choose Taiwan or HSK underlines without changing tone colours or pronunciation. TBCL, TOCFL and the different HSK editions are separate lists, not interchangeable rankings.

A word’s colour is its lowest listed level in the selected source. A missing word is unlisted, never automatically “advanced”. Whole words share an underline; characters keep their individual tone colours.

Personal imports remain in this browser and can coexist. The website operator must verify source accuracy and redistribution rights before installing a public list.

CSV: word,level,pinyin; pinyin optional. TBCL: 1–7, including source sublevels 1*, 2*, 3*, 4*. TOCFL: N1, N2, 1–6 or unsplit C. HSK2: 1–6. HSK3: 1–9 or grouped 7-9, exactly as sourced. Do not split grouped levels by guessing.

Restoring TBCL never removes HSK, or vice versa. Traditional/simplified aliases use OpenCC when installed. Original text and imported lists are preserved.

FIND YOUR READING VOICE

A little type tasting

Choose a card to change the reader immediately. Web fonts load from this website, only as needed. Device choices use installed fonts. Missing characters can still fall back.

Text tools & safekeeping

Recorded audio, precisely followed

Attach an audio file and its timing JSON. Timings must match the displayed text exactly. Playback follows the supplied timestamps. This is optional; browser narration needs no audio files.

No pair selected. Maximum audio file size: 50 MB.

Audio can be saved with a saved text. Library backups exclude audio files; retain the originals. Script conversion invalidates timing unless the displayed text still matches.


A little light on how it works

Reading & pronunciation

Tap a word for definitions, then tap a character in the word to inspect it. Pinyin is machine-assisted and can be wrong for names, ambiguous words, or Taiwan/Mainland differences. Correct a particular occurrence in the dictionary panel. Displayed tones are dictionary tones, not a full transcription of tone sandhi. Zhuyin is transliterated from the same pinyin, not a separate authority.

Flow mode uses the voice’s boundary events when available, otherwise one still passage marker shows the current position. Pause/Resume preserves the device voice’s current position. Speed changes apply to the next phrase without resubmitting it. Word-paced mode follows each spoken word’s start event. Character drill gives isolated sounds and may misread polyphonic characters. Narration uses the voice’s pronunciation, not your on-screen correction.

Your own text

Use Import for Chinese copied from a website, plain text files or PDFs with a selectable text layer. Review extracted PDF text before reading. Scans need external Chinese OCR first; the app does not reproduce PDF page layout. Display conversions always start from your saved source; Text tools can show or export that source.

Follow-along & tone paths

No tone cues switches off tone marks, shapes, colours and motion together. Plain pinyin and Characters only suppress extra shapes. Choose pronunciation in the toolbar; press P to hide and restore it. Voice, tone motion and appearance settings are inline beside your text, or below it on a phone. Natural flow speaks connected phrases; slower word and character drills are optional. Tones only uses large drawn contours. Pinyin + tones uses unmarked spelling with one drawn tone; small pronunciation marks are an alternative appearance. Tone paths show each dictionary tone’s shape above the character. Tone motion is optional. A timed character moves along its dictionary-tone contour. Every starter character has an audio position; some recovered positions are approximate. A frame follows reported words; uncertain passages have one still location marker. Gentle starter recordings and prepared narration are spoken at a slower pace before their audio is aligned. Live speed adjustments preserve pitch. Paths stay visible even without precise timestamps. Connected speech can change dictionary tones; these shapes are learning cues, not live pitch tracking.

Privacy & persistence

Normal reading needs no account and does not upload your documents. There is no tracking or analytics. AI practice story generation is disabled on this deployment. Selecting and saving vocabulary stays local. PDF bytes and opening passwords are not sent. Do not submit confidential material. Books, settings and vocabulary live in IndexedDB on this browser and origin. They are not encrypted or synchronized. Other people using this browser profile can access them. Browser eviction, private browsing, clearing site data or changing the website address can remove access. Export backups.

Text analysis is local. A voice marked “network” may send text to its operating-system or browser voice provider; consent is requested before using it. A “local” label is the platform’s own report, not an independent audit. Timed audio attached here plays locally. Optional MeloTTS neural reading sends only the reviewed page text to this website’s separately authenticated VPS service after consent. Once models are installed, that worker performs inference offline, without a speech-provider API. Generated audio is temporary in this tab; the server copy is deleted after download where possible and expires under the configured retention policy. Logical deletion does not erase snapshots or backups.

Keyboard & display

Space play/pause · Esc close a panel (or stop) · +/ character size. Tab to words and press Enter for the dictionary. The toolbar above the text holds playback, voice speed and four independent size controls: Characters (24–128 px), Pinyin or Zhuyin (12–72 px), Tones (16–80 px) and Website (100–180%). Size buttons stay under your pointer as settings change. During playback, Pause/Resume, Stop and speed stay at the top edge. Scrolling manually turns off Follow text; check it beside playback to resume following. Use Alt +/− to resize that layer and P to hide/show it without forgetting its size or mode. Hidden layers must be shown before resizing. Ctrl/Cmd browser zoom is not intercepted. Browse live font previews to choose a self-hosted family after website preparation, or use an installed-device stack. The selected preview reports loaded or fallback status. This site includes its own prepared font subsets. Tone colours belong to the Chinese characters; Taiwan/HSK underlines are independent.

Data & acknowledgements

Original Lantern sample texts and starter definitions are included. ToneWick uses CC-CEDICT (CC BY-SA 4.0 or 3.0 as declared by the downloaded snapshot; inspect the retained header), pinyin-pro and OpenCC. The installer preserves upstream licences. Taiwan and explicitly versioned HSK datasets remain separately attributed. No Du Chinese stories, audio, branding, or code are copied.

Colour palettes use analogous hues and restrained complementary accents. Research does not establish a universally pleasing “correct” palette. Colours have text labels and can be switched off.

Source links, data licences and design references →

THE STORY GARDEN · OPTIONAL AI

Same words. A new adventure.

Meet difficult words in another setting. Choose the challenge, keep the words you need, and let a new story grow.

1 · Choose your material

ComfortHardest here

2 · Shape your practice

Learning and saved words get priority. “Known” words are omitted from suggestions. Change the selection freely; a different challenge resets it.

0 selectedYour familiarity is saved locally

3 · Ask your story runner

Only Create sends this reviewed excerpt and selected words to your VPS and chosen AI provider. Normal reading remains local. Do not submit confidential material.

Target selection stays in your browser. AI generation needs an enabled VPS runner.

Recent requests · resume after reconnecting