Choosing what the audio reads: skip the boxes, the « read also » blocks and the rest
An article page holds far more than the article: pull-out boxes, « read also » blocks, subscription prompts. The « Text read » button in My audio lets you choose exactly what the voice reads, and here is what it changes about cost.
On an article page, the article itself is rarely alone. Around it sit a context box, a « read also » block, a subscription reminder, sometimes a photo caption or a repeated standfirst. A human reader skips all of that without thinking. A voice does not: it reads what it is given. If nothing is arranged, the audio can open with « Subscribe to read on » or close with three neighbouring headlines, which makes no sense to the ear. That is exactly the problem the « Text read » button in My audio solves.
What the « Text read » button does
From the My audio screen, every audio carries a « Text read » button. It opens a dedicated page that shows the text the voice reads for that audio. By default it is your article's text, the one pulled from your site or pasted at creation. You can replace it with the version you actually want to hear: a text stripped of the boxes and neighbouring blocks, or a version rewritten for listening outright (a narrative instead of a table, an acronym spelled out the first time, a caption turned into a sentence). You paste that text, you confirm, and the audio is redone with it.
Two limits frame that text, and they are deliberately simple. Below a readability floor (a few hundred characters), the text is judged too short to be voiced, and the button tells you so rather than producing a one-sentence audio. Above a hundred thousand characters, it is too long for a single reading, and the message invites you to split it into several audios. Between the two, you are free.
What the chosen text guarantees over time
A chosen text is not a setting that gets forgotten at the first change. It stays attached to the article, and that is the whole point. Later regenerations read it again, whether they start from the dashboard or the API. Above all, an update to the article on your site does not overwrite it: normally the page is read again before a regeneration, to catch a corrected article, but once you have set a text, reading the page again would undo your choice, so the tool refrains. If your CMS pushes the article over, the page is filed in the database and the audio does not move.
And if you want to go back, it is a single gesture: « Go back to the article's text ». Your site's page is then read again, and voiced as it is. That return, too, redoes the audio, so it costs (it is a regeneration), and the screen now works out the cost and shows it to you before launching. Nothing goes ahead without your having seen what it will cost.
What it changes about cost, said honestly
This is the concrete point. Having a chosen text read is taken from your volume like a regeneration: you pay for the characters actually read. The good news is that this maths works in your favour when you strip out the surplus. An article whose page carries a 400-character box, a 300-character « read also » block and a 200-character subscription reminder makes the tool read 900 extra characters at every generation. Remove them once in the chosen text, and you pay only for the article, and you stop paying for the rest at every later regeneration. The saving is not spectacular on a single audio, but it is real and it repeats.
The reverse is true too, and it is worth knowing: rewriting longer than the original costs more, since cost follows the number of characters read. The button is not a way to pay less, it is a way to pay for the right text. For a one-off fix (a word, a sentence) it is not the right tool either: our piece on fixing a sentence without regenerating describes the lighter gesture that stitches a passage back in without redoing the whole audio.
When to use it, and how it fits your site
« Text read » is built for the cases where the page and the listening diverge: a technical document that does not listen well as is, a page heavy with blocks, a format you want in a « told » version. For the normal flow of a site, the audio follows your content on its own, and a published correction carries through as we explain in updating a corrected article. Choosing the read text goes through the same regeneration route as everything else, which means it respects your plan, your volume and your voice: nothing else changes. If you are adding audio to a site, the page on WordPress and article audio shows where all this comes in, and « Text read » stays available article by article, alongside it.
At bottom, it is the same idea as pasting a text to generate audio without a URL, but tied to an existing article: you keep your site's page, and you decide separately what the voice makes of it.
Give your articles a voice with WeDispatch
This blog is itself voiced by WeDispatch. Curious how it sounds on your content?
Book a demo