WordPress: what happens to the audio when you edit a published article?
A published article gets corrected, and the question lands at once: what about the audio version, which does not move on its own? The exact regeneration path, what plays while it happens, and the real delay, described with no grey areas.
You publish an article, it gets voiced, all is well. Then you spot a typo, or add a paragraph, or a fact changes. You fix the text in WordPress. And there, a fair question: the audio version did not fix itself. What becomes of it? Here is the exact path, step by step, with no grey areas.
While you correct it, the online audio keeps playing
First point, and it is the reassuring one: the moment you edit the text, the audio already online keeps being served normally. A reader landing on the article while you correct it hears no gap, hits no error, finds no broken player. The old version stays playable until the new one is ready.
What the plugin does is mark the item "to regenerate." The text has changed, the audio no longer quite matches, and the plugin's panel in the editor says so clearly: it distinguishes audio ready, preparation in progress, and a scheduled go-live. So you know at a glance that the audio version is one correction behind, without anything breaking on the visitor's side.
Nothing is spent until you decide
Second point, and it avoids a common waste: fixing a typo does not spend a generation. As long as you have not relaunched the build, no credit is spent. This is a deliberate choice, because an article is often corrected in several passes the same day: a comma, then a word, then a subhead. Regenerating the audio on every save would burn a generation for nothing each time.
The good practice is therefore simple: let the text settle, make all your corrections, and only relaunch the audio once, when the written version will not move again. The plugin waits for your go-ahead rather than chasing every keystroke.
The regeneration gesture, and its delay
Third point: how you relaunch, and how long it takes. Two paths, depending on your setting.
In Manual, which is the default behaviour, you click "Generate audio" from the editor once the text is fixed. The plugin sends the new content back to the service, the audio is rebuilt, and generation is a matter of seconds for a typical blog post, not minutes. You do not wait around or reload ten times: the panel moves from "preparation in progress" to "audio ready."
In Automatic, there is a setting that regenerates on its own when the text of an already-voiced item changes. It is deliberately off by default, precisely because each regeneration uses one generation: you turn it on only if you prefer systematic freshness to control over the counter. It is a trade-off you set once, per site, knowingly.
Why the player points at the right version on its own
Fourth point, the quietest and the most important: once the audio is regenerated, you have nothing to redo in the article. No link to update, no id to paste back. This works because the player never freezes a reference to a file at the moment you save: the block or shortcode you placed does not memorise the audio id, it resolves the current version at display. A regenerated article changes id on the service side, and the player follows without you touching it. This insertion mechanism, described in detail in add audio to a WordPress article without touching the theme, is what keeps a correction from ever leaving a reader pointed at a vanished audio. The WordPress plugin handles that wiring for you.
Two neighbouring cases that do not ask for the same thing
You have to distinguish, because "correcting an article" covers very different gestures.
Correcting the meaning or the text (a wrong sentence, an added paragraph) requires regeneration: the audio must say what the article now says. That is the case above.
Correcting a pronunciation (a mangled proper noun, a misread place name) is a lighter and often free case. There you do not redo the whole article: targeted correction reworks only the passage concerned, which requires knowing exactly where it sits in the audio. And if you only wanted to pull a spoken quote, the cut reuses the existing audio with no regeneration and no cost. Confusing these three gestures means regenerating where a targeted correction was enough, or the reverse.
In short, the whole path
You correct the text: the old audio keeps playing and the item moves to "to regenerate," spending nothing. You let the text settle. You relaunch, by hand or automatically depending on your setting, and the audio rebuilds in a few seconds. The player points at the new version on its own, because it never memorised the old one. No step asks you to touch the theme, and none leaves a visitor in front of a silent player.
Give your articles a voice with WeDispatch
This blog is itself voiced by WeDispatch. Curious how it sounds on your content?
Book a demo