WeDispatch 1.5.4: settings that keep their word

There are two ways to disappoint a customer without ever lying to them. The first is to let them switch on a feature that will produce nothing. The second is to show them information that was accurate yesterday and false tomorrow, because the product moved underneath it. This version addresses both, and does so where it counts: in the structure, not in the display.

Highlighted text now follows what the reading can actually produce

Text that lights up as you listen is not a visual effect: it follows the reading word by word. It therefore needs, inside the audio, the position of every word — a piece of data that not every reading quality returns. Four features rest on that same positioning: synchronised text, correcting a passage without regenerating everything, aligned subtitles, and cutting a quote to the second.

Until now, that dependency lived in our heads. It now lives in the product. Wherever the selected reading does not provide the positioning, the switch is disabled and the reason is written right beside it — on the account, on each site, and on the accessibility page. Nobody ticks a box only to find out three days later, through a listener, that it produced nothing.

Changing tier warns you before it writes

A change of reading quality that would cost you the highlighting no longer applies in one click. A screen sets out what would be lost, makes clear that audio already produced keeps its alignment and stays highlighted in the player, then asks for an explicit confirmation. If the setting was on, it is switched off at the same time — and the screen says so. A setting that changes without announcement is always discovered at the worst moment, and rarely by the right person.

Going back requires no catching up: returning to a tier that produces the positioning restores the feature as it was, including on sites that had set it on their own side.

The constraint is attached to the reading, not to a tier name

This is the point we care about most. We could have written, somewhere, "this tier has no highlighting". It would have worked — right up to the day that tier gains word-level positioning, and the restriction stayed in place with nobody left to remember why. So the rule is derived from what the reading actually returns. The day a voice engine gains that capability, the feature reopens by itself, with no intervention and no possible oversight. An automated check now forbids the shortcut.

An audio's provenance is no longer inferred second-hand

The tier shown on an audio and the place where its voice was processed now come from a single explicit table, rather than being reconstructed by inference. This is a substantive guarantee: a provenance that named the wrong continent would betray the very promise it exists to attest. The fallback stays cautious — no European location is ever claimed by default, only when it is true.

In the same move, every message a customer reads now speaks in reading tiers — standard, premium, sovereign — and an automated check sweeps those surfaces on every release so it stays that way.

What this changes for you today

For the vast majority of accounts: nothing visible, and that is deliberate. Current reading qualities produce the necessary positioning, and highlighting works exactly as before. These safeguards exist for the day the choice widens — and that day is close.

We are preparing the arrival of a voice engine processed within the European Union for the sovereign tier. This version is what allows us to welcome it without any setting starting to lie along the way.

Other release notes

• WeDispatch 1.5.3: every blog article now has its audio version
• WeDispatch 1.5.2: the status page now watches text preparation too
• WeDispatch 1.5.0: European preparation now covers the whole chain