Accessibility: Read Aloud

The Read Aloud module turns the content of a page of your PDF into speech, using the macOS text-to-speech voices. It is designed to make it easier to consult a document alongside another activity, or to assist people with a visual impairment. You reach it from the Reading category of the Dashboard.

How it works

Reading applies to the current page of the document: HtkPDF extracts the text of the page currently displayed, shows it in an editable preview area, then reads it aloud with the selected voice. To read another page, simply navigate to it; reading in progress stops automatically whenever the page changes, and the associated text is reset.

The application uses the native macOS speech synthesis engine (AVSpeechSynthesizer). The available voices are therefore the ones installed on your Mac. You can add or download more from System Settings > Accessibility > Spoken Content.

Playback controls

The upper panel gathers the transport buttons:

Play / Pause: a large button whose icon changes with the state. On the first click it extracts the text and starts reading; while reading it becomes Pause; when paused it resumes reading where it left off.
Stop: definitively interrupts the reading in progress and resets the state. The button is disabled when speech synthesis is not active.

Speech rate

A Rate slider adjusts the speaking pace. The value shown (for example "1.0x") is an indicative multiplier: the minimum and maximum range allowed matches the one macOS offers for its speech synthesis engine. The adjustment applies from the next reading onwards (press Play again if needed to hear the new rate).

Choosing the voice

The Voice menu lists every voice installed on the system, along with its language code (e.g. Thomas (fr-FR), Samantha (en-US)).

The Automatic option (offered at the top of the list) selects the voice matching the current macOS language. If no voice matches the system language exactly, HtkPDF uses the default voice associated with the current language code.

Using OCR

The Use OCR switch specifies the text extraction method used before reading:

Off (default): HtkPDF reads the vector text layer embedded in the PDF. This is the option to favour for natively digital documents.
On: if the page contains no text layer (scanned PDF, vectorised image), HtkPDF runs optical character recognition on the page image. The result can take a few seconds, indicated by a "Extracting text…" indicator.

OCR is only useful for scanned documents; on native PDFs it needlessly lengthens preparation time.

The text preview

The Text being read area displays the extracted text before it is read. This area is editable:

you can correct extraction errors (misrecognized characters, unwanted line breaks);
you can shorten the text to read only an excerpt;
you can also paste your own content into it to have it read — the feature then acts as a text reader independent of the PDF.

When you press Play again while the preview area already contains text (from extraction or from your own typing), HtkPDF reads that content directly, without re-running page extraction.

If extraction returns nothing (for example on a completely blank page, or an image with OCR disabled), the message "No text found on this page" is shown instead.

Best practices

To go through a long document aloud, navigate page by page between readings: the module does not read the PDF continuously.
If the automatic voice does not match the document language, select a voice in the right language manually — pronunciation will improve markedly.
If you notice segmentation errors (words run together, line ends read as full stops), correct the text directly in the preview before starting reading again.

Note: speech synthesis stops automatically when you leave the module or change page. To avoid an interruption during a long reading, do not trigger navigation or editing actions in parallel.

Screenshots

1 screenshot — click to enlarge