Pages that turn themselves, in time with the reading.
A read-along is a stack of pages and a recording of someone reading them. This is the one workflow ClapperBoard can time for you: it listens to the narration, reads the pages, matches one against the other, and sets every page's duration to fit.
Words in the audio. The match is made between what is spoken and what is printed, so the recording has to be speech. Music has nothing to match against — for a score, see sheet music, where the timing is placed by hand.
Readable text on the pages. A PDF with a real text layer is ideal; ClapperBoard uses it directly. Scans and photographs are read with text recognition instead, which works well on clean pages and less well on ornate type, heavy layout or poor contrast.
Pages and a recording in the project. Tools › Sync to Audio stays unavailable until there is at least one image, and at least one audio item on the lane the sync will read.
Choose Tools › Sync to Audio. What follows is one sheet that moves through several stages.
First it asks which audio to handle, given as a range — From and To — and tells you how many items there are and how many are not yet transcribed.
A project with more than one audio lane also gets an Audio lane picker here. The sync reads one lane, and it starts on the topmost one on the assumption that the narration lives there — deliberately not whichever lane you last clicked, so a long transcription run cannot be aimed at a music bed by accident. Pick another and the list of audio items below redraws to that lane's.
Transcripts are kept, so a second run does not redo work already done. On a long book split into many audio files, that means you can transcribe a few at a time rather than committing to the whole thing in one sitting.
Alignment is the slow part on a long book, and it shows a progress bar rather than a spinner because it can take a while.
Alignment quality can be changed here, and again in the review. Default compares everything. The faster settings skip common words, or compact long pages, or both — each trades a little accuracy for speed. It is worth starting with Default and only reaching for the others if a book is large enough to be tedious.
Nothing changes until you agree to it. The review lists every page with:
| Column | Meaning |
|---|---|
| Before / After | The page's current duration, and what the sync proposes. |
| Confidence | How well that page's words matched the narration. |
| Skip | Leaves this page's timing alone. |
Low confidence is a signal to look, not a failure in itself — an illustration-only page has few words to match, and will score low while being perfectly well placed by its neighbours.
Skip is for pages that should not be retimed at all. ClapperBoard suggests it for pages that look like a table of contents or an index: a page of chapter titles and page numbers has plenty of words, none of which are read aloud, so it would otherwise attract matches it should not have. Reset Skips clears your choices and starts again.
The review also lists the gaps it found — real silences in the narration, which are usually where one page ends and the next begins.
Rather than cutting a page at the exact word the alignment matched, ClapperBoard centres each break in the largest genuine silence nearby, so a page does not turn in the middle of a spoken phrase. The break offset lets you shift that.
If the pages run past the end of the recording, the review says so and offers to add silence to the end of that lane, so the two end together instead of the video continuing over nothing.
Apply Changes commits the durations. Everything is one undoable step, so a sync you dislike can be undone in one go.
The sync may add spacer items to keep the lanes aligned. Reorder mode removes them, because freely reordering items would invalidate their positions — ClapperBoard asks before doing it. Nothing is lost: running the sync again regenerates them.
If the book has a table of contents, ClapperBoard reads it during the sync and turns it into a Page chapters button in the toolbar, listing the entries with their printed page numbers. It is a quick way to jump to a chapter while checking the result. A table of contents set in columns, with the titles and the page numbers ranged apart, is read as well as a plain one.
The button appears only when a table of contents was found, and re-populates each time you apply a sync. What was found is kept with the project, so the button is still there when you close the document and open it again — you do not have to re-run the sync to get it back.
An audio file that carries chapter markers of its own gets a second button beside this one, and those markers can also be cut into separate items. More on chapter markers.
Whole pages are a coarse unit — a page of a novel can hold a minute of narration. Smart Text Crop cuts each page into separate items, by paragraph, by a set number of blocks, or every few lines, and each becomes its own item on the timeline.
Do that before syncing. The sync then times those smaller pieces, and the video follows the reading a paragraph — or a line — at a time.
A single long recording can be cut where you want it: put the playhead where the split belongs and press S — Split Item at Playhead. Useful for dividing a book into chapters so they can be transcribed a few at a time. It cuts the item on whichever lane is selected, so with several lanes select the recording's lane first.
Audio items can also be faded, and their levels shaped — see transitions. The level of a whole lane is set in the lane controls, which is where to put a music bed under a narration rather than fading it clip by clip.