An article's strongest line, as a share-ready audio excerpt

The best line in an article is almost always buried in the middle. Eight minutes of audio, and the line that lands arrives around the fifth. To make it travel on social media, you used to have to re-record the sentence, or open an audio editor and trim the file by hand, or give up. Most newsrooms give up, and their best content never leaves the page. That's waste, and it's avoidable.

Cut, don't regenerate

The whole idea rests on one distinction. When an article is voiced, WeDispatch keeps more than the audio file: it also keeps, word by word, the moment each word is spoken. We described that timing in detail in our article on word-level timestamps, and it's what changes everything here. To pull out a quote, you don't make new audio: you cut the existing audio, exactly where the sentence starts and ends.

From the newsroom side, it's two clicks. In your account, a menu lists the articles you've already voiced. You open the transcript, click the first word of the sentence, then the last. The selection highlights, the duration shows live, and you download the MP3. That's it. No editing suite, no levels to set, no export settings to configure.

Why the cut is clean

A cut made by ear always damages something: a word clipped at the start, a syllable lost at the end. Here the cut relies on the timestamps, not on hearing. It starts 150 milliseconds before the first syllable and leaves 250 milliseconds of breathing room after the last word. The result: never a truncated word, never an abrupt ending. The sentence begins and ends as if it had been recorded on its own.

And because you're reusing audio that's already been produced, the cut costs nothing. No new call to the synthesis engine, no characters counted, no credit spent. You can pull three versions from one article, or one a day for a week, without any of it touching your quota. The only limit is length: beyond three minutes it's no longer a quote, it's the article, and the interface tells you so before export.

What it's actually for

A short excerpt, in your brand's voice, drops in anywhere: a post on LinkedIn or X, a story, a newsletter, a teaser message. It does two things at once. It lets people hear the tone of an article, which a headline and a thumbnail never will, and it brings the listener back to the full version, because a good line makes you want what follows. It's the natural extension of what chapters and timestamped links already allow: there, you share an entry point into the article; here, you detach a self-contained piece of it.

For a community manager with no studio, for a newsroom that wants to feed its channels without losing a day to it, for a publisher monetising attention, the audio quote turns content you've already made into material to circulate. The synthesis work is done; all that's left is choosing the right sentence.

What it takes to work

The feature rests entirely on word-level timing. It's therefore available on articles voiced since that timing arrived, in mid-July 2026. For older audio, produced before, a simple regeneration makes it eligible: the article runs through the engine again, its timestamps are captured, and the cut becomes possible. The rest of how the voice works, from choosing a timbre to reading French, is covered on our French text-to-speech page.

What we don't do, and why

We could add trimmings: a jingle, a music bed, an animated card with scrolling text. We don't, at least not here. An audio quote should stay what it is, an exact piece of the article, in the same voice, with nothing added that would suggest a different production. The day the line travels, what matters is that it sounds like the article it came from, so the listener who clicks finds the very voice they heard. The cut is faithful, free and instant: that's already a lot, and above all it's what makes it dependable.

Read next

• An audio player in the page or in an iframe: what that choice decides
• Image captions and photo credits: what French text to speech does with them
• Footnotes and note markers: what French text to speech does with them