Help Center

Enter a session title, choose and preview a script, choose a follow mode, and select terminology if needed. The session uses the built-in speech model automatically. Open without saving or save and open, select a microphone, then choose Start session. Preparation must be complete and the built-in model and device available.

Markdown (.md, .markdown), plain text (.txt), Word (.docx), and unencrypted text-based PDF (.pdf) are supported. Scanned and image-only PDFs are not supported. Save legacy .doc files as .docx first. PDFs import as text without the original layout; check the content and order in the preview. Empty scripts, documents without readable text, or scripts with no matchable Chinese or English body text show a message and cannot be saved. Add body text and select the script again. Your original file stays unchanged.

NoteDOCX imports body text and converts recognized headings and lists; it does not preserve the full Word layout. Check the preview.

Check microphone permission, the device connection, and the selected input device. Confirm that the model is available before starting. If a device or model fails, stop the session, resolve the displayed issue, and select an available device.

During silence, off-script explanations, insufficient recognition evidence, or repeated or similar content whose position cannot be safely distinguished, the app keeps the confirmed position to avoid incorrect advances. Continue with matching script content, or reposition manually using the controls supported by your display mode. Repeated content does not always prevent Sequential following; the position must still meet the confirmation conditions.

Open without saving does not create a saved record. Save only stores the session locally without opening the reader. Save & open stores the session and opens the reader. Speech following starts only after you select a microphone and start the session.

No recording is required. Entries contain a canonical writing, other valid writings, a category, optional notes, and an enabled state. Other writings are valid text forms of the same term in a script, not recognition errors, pinyin, or pronunciation mappings.

The floating teleprompter shows the script in a separate window with always-on-top and background opacity settings. At 0% opacity, the background is fully transparent when the controls are hidden; script text and highlighting remain. During interaction, a minimum background opacity keeps controls usable. Controls and the frame appear when the pointer enters and usually hide when it leaves. With the window focused, Tab also reveals the controls for keyboard use, during which they may stay visible. Open it from a running session or preview it in settings.

Version 0.1.0 uses the built-in primary and fixed auxiliary speech models delivered with the app. Optional model installation and model selection are unavailable. New sessions use the built-in primary model automatically. Speech following supports Chinese and English. Content detected as another language stays in the script and can be read in full-script displays, but does not participate in positioning. Focus playback shows only matchable units. Mixed-language scripts with matchable Chinese or English body text can be saved under the existing preparation rules without removing other-language content.

Add and enable entries on the Terminology page, then select them in the session form. Prepared and saved sessions use their stored terminology snapshot. Editing, disabling, or deleting library entries does not immediately change those sessions. To update a session, apply the terminology selection again while editing and save it.

Sequential matches the current unit and the immediately following unit to advance in order. Forward matches from the current unit to the end of the script and allows skipping later content. Free matches before or after the current position and allows returning to earlier content. In a newly opened session without manual positioning, Sequential starts at the first matchable unit. Sequential is the default. Follow modes control the matching range, independently of display modes.

When script analysis finds exact duplicate matchable body units, or a whole unit contained in another at positions allowed by complete-word and terminology boundaries, only Sequential is supported, not Forward or Free. Follow the analysis message. Preparing a Forward or Free session also requires a quick follow speed check on the current device. If it fails, switch to Sequential as prompted and try again. The app does not change your follow mode automatically.

In Default and Fixed displays, click a matchable body unit to reposition. Focus also shows the full script and supports clicking before the session starts. After starting, use the scroll wheel to switch units; click-to-position is unavailable. Manual positioning is unavailable in every display mode while paused; resume the session first. Following continues from the accepted position.

Default shows the full script and advances when the next unit reaches the end of the page or cannot be fully displayed, moving the current unit to the top. Fixed shows the full script and places the current and next units near the middle. Long content is positioned according to the available space. Manual positioning and speech following use the same display rules.

Before starting a session, Focus shows the full script and supports clicking matchable body text to reposition. After starting, it shows only the current and next matchable units, uses the scroll wheel to switch units, and hides images and other content that does not participate in following. Switching units is unavailable while paused. A brief scroll-wheel hint appears the first time Focus playback is entered after opening a session.

In Settings → Display & language, adjust text size, line height, and colors for completed, current, next, and unread content. The preview area demonstrates all three display modes and can open a floating preview. Settings are saved automatically on your device and used by the session reader and floating window. If saving fails, a message asks you to retry.

No. The primary and fixed auxiliary speech models are delivered with the app and need no download during use. Missing resources or loading failures prevent the session from running. Check the displayed message; the app does not download or repair resources automatically.

Pause temporarily stops audio capture. Resume continues from the paused position without processing speech from the pause. Stop ends the session; you can then start again or return to session management. The global pause / resume shortcut is enabled by default: Ctrl + F9 on Windows and Command + F9 on macOS. Some keyboards also need Fn. Change or disable it in Settings → Other settings → Session shortcut. When enabled and successfully registered, it pauses a running session or resumes a paused one, even while another app has focus. It does not switch states before starting, after stopping, while frozen by a fault, or while an operation is busy, and cannot start or stop a session.

Check the displayed message and confirm that the built-in speech model is available. Invalid records or damaged artifacts are rejected. You can delete unusable records from Saved sessions and create a new session from your original script. Before deleting, make sure your original script is still available.

Speech recognition, script matching, and following run on your device. Built-in models work offline. Opening the official website requires a network connection and does not automatically upload speech, scripts, or terminology.

Script copies, saved sessions, terminology, models, and reading settings are stored locally. When reporting a problem, start with the version, operating system, steps, and error message; avoid sending original scripts, recordings, or recognition content.

Help & updates shows the current app version and provides a link to the official website. See the website for release information. The app does not check for or install updates automatically.

NoteDownloads and updates for store-installed copies are managed by the relevant store. Automatic updates depend on store and system settings.