WeDispatch Audio Text to Speech
| 开发者 | toutestfaisable |
|---|---|
| 更新时间 | 2026年9月24日 06:12 |
| PHP版本: | 7.4 及以上 |
| WordPress版本: | 7.0 |
| 版权: | GPLv2 or later |
| 版权网址: | 版权信息 |
详情介绍:
Your readers get an audio version of every article. Your writers do nothing: they publish, and the
audio appears with the text. It is free to start, no credit card, and it takes about two minutes to
connect a site.
Try it in two minutes
- Install and activate the plugin.
- WeDispatch, then "Connect this site to WeDispatch". Your account is created on the way, and the key is stored here for you. Nothing to copy.
- On the overview, click "Read it aloud": your latest post that fits your trial is voiced first.
- Open the page: the player is there, and your article is read aloud.
- Happy with it? Switch to automatic, and every new post gets its audio on publication.
- Newsrooms. The audio is ready when the article goes live, including for scheduled pieces, and nobody has to remember a separate step on a busy day.
- Sites with an accessibility duty. Public bodies and large companies covered by the European Accessibility Act, EN 301 549 or RGAA: this removes one specific barrier, properly, and says honestly what it does not do.
- Shops and content sites. Product descriptions and long guides get an audio version, which is read where a screen cannot be looked at.
- Nobody has to remember anything. Audio is produced on publication, not by a separate step someone forgets on a busy day.
- Scheduled content is ready on time. An article scheduled for 6 a.m. has its audio prepared in advance and published at the same moment as the text. The first reader of the day finds audio, not a "preparing" message.
- A voice per section and per author. Sport does not have to sound like the obituaries. The routing is category, then author, then site default, and the editor panel shows which voice will actually be used, before publication.
- Correcting a typo does not spend a generation. When the text of something that already has audio changes, the audio online keeps being served and the item is marked "to regenerate". You rebuild it once, when the text has settled.
- Subscriber-only audio, with nothing to code. If your full articles are reserved for subscribers, the audio follows the same rule. Your site attests to a status ("subscriber"), never to an identity, and no personal data about your readers leaves your server.
- The player is exposed as a labelled landmark, so a screen reader announces it and its user can jump straight to it instead of hunting through the article.
- The player is keyboard operable and does not start on its own.
- The audio version is an alternative to the text, never a replacement for it: the written article stays complete and indexable.
- Two modes: Auto (audio is generated on publication) or Manual (the author generates it from the editor, then publishes).
- Any content type: posts, pages, products, and any custom post type registered by your theme or your plugins. You tick the ones you want.
- Automatic voice routing: category, then author, then site voice, with a preview, inside the editor, of the voice that will actually be used.
- Multilingual reading: French, English, Spanish, German, Italian, Hindi, Chinese, Arabic, Portuguese, Japanese, Hebrew, Korean, Dutch, Swedish, Polish, Tamil, Telugu, Russian, Turkish, Thai, Filipino, Czech and Finnish, with a native voice per language. The language of each item is detected automatically (Polylang, WPML, or the site language). Each language is offered on an account once its voices have passed human review.
- Scheduled content: the audio is prepared ahead of time and goes live at the same moment as the text.
- The player goes at the top or the bottom of the article, or exactly where you want it: an editor block, a widget, or the
[wedispatch]shortcode. - Optional synchronised text (word-by-word highlighting during playback).
- An optional transcript under the article, folded, for readers who cannot use audio and for search engines.
- An optional "Listen next" list, based on the article being read and never on who is reading it.
- A "My subscription" panel in the settings: plan, monthly usage and a shortcut to subscribe.
- Loaded through the standard WordPress script queue, deferred and in the footer, so caching and performance plugins handle it like any other asset.
[wedispatch] shortcode, so it drops into any builder that
accepts one of the three.
- Page builders: Elementor, Divi and WPBakery, through the block, the widget or the shortcode.
- WooCommerce: product descriptions are read aloud like any other content type; tick "Products" in Settings.
- Multilingual: Polylang and WPML. The language of each item is detected, and read by a native voice.
- SEO: Yoast SEO and Rank Math. The optional transcript is plain text under the article and does not alter your schema.
- Caching and CDN: WP Rocket, LiteSpeed Cache and Cloudflare. The script is queued the standard WordPress way, deferred and in the footer, so it is handled like any other asset.
- WordPress.com: on plans that allow plugins.
安装:
- Install the plugin (Plugins, then Add New, then Upload Plugin) and activate it.
- Open the WeDispatch menu in the sidebar and click "Connect this site to WeDispatch". You sign in or create your account on wedispatch.fr, you authorise this site, and the key is created and stored here for you. There is nothing to copy, and your password never enters WordPress.
- WeDispatch, then Settings: choose the mode (Auto or Manual), tick the content types to read aloud, and pick the player position.
- WeDispatch, then Voices: pick the site voice, and if you want, a voice per category and per author.
- Publish something: its audio version is generated and the player appears.
屏幕截图:
常见问题:
Is there a free plan?
Yes. Your first articles are on us, so you can hear your own content before deciding anything. Paid plans start when you need volume, and they are listed on wedispatch.fr. The plugin reads what is left from your account, so what you see in WordPress is always your real allowance.
Do I need a credit card to try it?
No. You create the account, connect the site and publish. A payment method is only asked for when you choose a paid plan.
Does it work with Elementor, Divi or WPBakery?
Yes. The player is available as an editor block, as a widget and as the [wedispatch] shortcode,
and every one of those three builders accepts at least one of them. You can also leave the player at
the top or the bottom of the article and not place anything by hand.
Does it work with WooCommerce?
Yes. Products are a content type like any other: tick "Products" in Settings and product descriptions get an audio version when they are published.
Does the plugin work without a WeDispatch account?
No. It is the connector for an online service: the site has to be connected to a WeDispatch account. Until then, nothing is read aloud and no request is made on its own. Two things happen only when you click: connecting the site, and, on the WeDispatch overview, a button that lets you hear 20 seconds of your latest post, read by the service, before creating an account.
The player is not showing up on my page. What now?
Open the Contents screen and use "Check the display" next to the item. The site will look at its
own public page and tell you what it finds. The usual cause is a theme or a page builder that
renders the content through its own machinery, so no plugin can insert anything into it. In that
case, add the WeDispatch block where you want the player, or paste the [wedispatch] shortcode:
both work regardless of the theme.
If the check answers that the page could not be read, that is not a verdict on the player. Some
hosts block a site from calling itself. Open the page yourself in a private window instead.
Does it work on WordPress.com?
Yes, on plans that allow installing plugins. Two things behave differently there and are handled: the site identifies itself to the service under a different name, and calls from the site to itself may be refused, which only means "Check the display" cannot look. It says so rather than blaming the player.
What data leaves my site?
The text of the item, its title, its URL and its ID, at the moment of generation. See the "External services" section above, which lists every request and what triggers it.
Can it read pages and custom post types, not just posts?
Yes. The settings screen lists every public content type on the site (posts, pages, products, and anything a theme or another plugin registers) and you tick the ones that should be read aloud. Posts are the default.
Can I choose the voice per author or per category?
Yes. In Auto mode, the voice is chosen in this order: category, then author, then the site default voice. The editor panel shows which one will be used.
What happens if I edit something that already has audio?
The audio that is online keeps being served and the item is marked "to regenerate": fixing a typo does not trigger a new reading. Once the text has settled, the "Generate audio" button in the editor panel produces the new version. A setting also lets the plugin handle that itself; it is off by default, since each regeneration uses one generation from the plan.
My site is multilingual: which language is each item read in?
Its own. The plugin reads the language from Polylang or WPML, and falls back to the site language
if no multilingual plugin is installed. Voice lists are grouped by language, with the language of
the item first. To impose a different rule (custom field, language per category), use the
wedispatch_post_language filter.
What if I schedule something for tomorrow morning?
The audio is prepared as soon as the item is scheduled and goes live at the planned time, together with the text. The first reader of the day finds the audio, not a "preparing" message.
Where does the player go?
At the top or the bottom of the article by default. Set the position to "Manual" and place it
yourself: the WeDispatch audio player block in the editor, the widget of the same name in a
sidebar, or the [wedispatch] shortcode anywhere a shortcode works, including a theme template or
a page builder.
Is the plugin's own interface available in my language?
Every screen is translatable and a French translation ships with the plugin. Other languages can be contributed on translate.wordpress.org, and WordPress installs them automatically once they are published. Contributions are very welcome, and they should be uploaded by the translator from their own WordPress.org account so that the locale team can review them with the right person.
Does the plugin slow my site down?
No. Generation happens on the service side, off your server. The player is a small script loaded through the standard WordPress queue, deferred and in the footer, so caching and performance plugins handle it like any other asset.
I was using an earlier version installed by hand. What do I do?
Version 1.10.0 changed the plugin folder name. Deactivate the old "WeDispatch" plugin, install this one, and activate it. Every setting, every voice and every audio version already produced is kept: they are stored in the database, not in the plugin folder. The old plugin can then be deleted.
更新日志:
1.31.0
- New: one audio per notice. For a post made of several notices, such as a daily chronicle or a list of dates, tick "One audio per notice" in the editor panel: each block (or each heading of the same level) gets its own audio, titled after the notice, and its own player right under it, outside the notice's coloured block. A notice whose text did not change is not charged again.
- New: a changed post no longer means paying for the whole audio again. When the text of a post that already has audio changes, "Generate audio" now shows what changed and two prices before anything is spent: read only the changed passages again, stitched into the existing audio at the price of an edit, or regenerate everything. Nothing is charged until you click.
- Changed: automatic regeneration reads only what changed when it can. With "Regenerate automatically" on, the service now reads again only the changed sentences when a few of them changed, and still redoes the whole audio when the changes are too large, the voice changed, or the post is scheduled.
- If the comparison cannot be made, the editor says why and offers the full regeneration, with its price when the service gives one. It never starts it on its own.
- New: the plugin tells you where to start. Right after activation you land on the WeDispatch overview, the plugin list has a "Connect" link, and until the site is connected a notice on the dashboard and the Plugins screen offers the one-click connection. Each administrator can close it for good.
- New: after connecting, you land on the overview, not on the settings. It offers to read one post aloud straight away: during the trial, the most recent post that fits what the trial has left, so the first audio never fails for being too long.
- New: hear your own post before creating an account. On a site that is not connected yet, the overview offers a 20-second sample of your latest post. It is sent only when you click, needs no account, and the text is not kept.
- Changed: new sites start in manual mode. Nothing is read aloud on its own until you decide, and once a first audio exists the overview offers automatic mode in one click. Sites already connected keep the mode they had: nothing changes for them on update.
- New: a post refused for lack of credits says so. The service's message now shows on the post's line in My content & audio, and a notice on the dashboard gives the way out. Until now these posts simply read "No audio".
- The trial is no longer described as "lost at the end of the month": trial credits do not expire.
- The overview no longer asks you to copy an API key: connecting the site creates it for you.
- New: the plugin asks for its own translation, from the people who can do it. When WordPress runs in a language the plugin does not speak yet, a notice on the WeDispatch screens, and nowhere else, says so and links to the translation project. The state is measured on the text domain actually loaded, not guessed from a file name, and the language is named in its own language whenever the server can do it. The notice carries the instruction that matters: upload from your own WordPress.org account, so that the locale team can give feedback to the person who did the work. "Do not tell me again" is final, and it holds for one administrator without silencing the notice for a colleague.
- The readme carries the same call, and the same instruction.
- New: eighteen more reading languages. Hindi, Chinese, Arabic, Portuguese, Japanese, Hebrew, Korean, Dutch, Swedish, Polish, Tamil, Telugu, Russian, Turkish, Thai, Filipino, Czech and Finnish join French, English, Spanish, German and Italian. This plugin already knows every one of them: automatic detection (Polylang, WPML or the site language), voice per category and per author, translated editions. Each new language becomes selectable on your account once its voices have passed human review, so they open one by one rather than all on the same day. Nothing to configure if you do not publish in these languages.
- New: fix a pronunciation without leaving WordPress. On the Voices screen, write a word as it is written and as it should be said. The correction lives on your WeDispatch account, applies to every audio generated from now on, and the screen says plainly that already generated audio keeps its sound until rebuilt.
- New: your podcast, finally visible. Every site with audio has had an RSS podcast feed all along; nothing showed it. The overview now gives the address and the submission links for Spotify and Apple Podcasts. Submit once; platforms refresh the feed on their own.
- New: listens next to the content that earns them. The Contents screen shows a Listens column, from the same 15-minute cache as the Statistics screen. If the figures cannot be read, the column stays away rather than showing zeros that would read as "nobody listens".
- New: a map of what lives where. The overview ends with two columns: what you do here in WordPress, and what lives in your WeDispatch account, each entry a direct link. The product lives in two places and no screen said the boundary.
- The onboarding checklist gains a fifth step, "Player seen on the public page", ticked by the display check itself, on facts. The four previous steps could all be green while visitors got a silent page.
- The admin stylesheet now uses logical CSS properties throughout, so right-to-left languages (Arabic among them) get a correctly mirrored interface for free.
- New: "Check the display". On the Contents screen, next to each item that has audio, a link asks the site to look at its own public page and report whether the player is actually there. "Audio ready" and "player on the page" are two different things, and until now nothing told them apart: a theme or a page builder that renders content without going through WordPress leaves the audio ready and the page silent. The check names that case and points to the block or the shortcode, which work regardless of the theme.
- New: a notice when some of your content is never read aloud. Only Posts are enabled by default. If your site has Pages, Products or any other public type with published items, the plugin now says so on its own screens, with the number, and a link to the setting. It appears only on WeDispatch screens, and "Do not tell me again" is final.
- New: one optional question when you deactivate. Nothing is sent unless you click "Send". Deactivating without answering is a button of its own, and pressing Escape or clicking outside does the same. What would be sent is listed in the window itself: the reason you pick, your comment, and the plugin version. Never any content, never anything about your visitors.
- Sites hosted on WordPress.com are recognised as WordPress sites by the service. They identify themselves differently from a self-hosted install, and were being filed under "API", which made a perfectly working plugin look unconnected in the account area.
- The Custom CSS box is gone from the Player screen. The plugin no longer takes a stylesheet typed into WordPress, in line with the directory guidelines. The setting itself has not changed and is still in your WeDispatch account, where it applies to this site exactly as it did before; the Player screen links straight to it.
- Your plan is counted in credits, and the screens now say so. An item costs one credit per thousand characters started, so a long read costs several: calling them "items" was a shortcut that stopped being true.
- The installation steps no longer tell you to copy an API key by hand. Connecting the site creates its key for you, from the WeDispatch menu, and your password never enters WordPress. Pasting a key still works for those who prefer it.
- Every script the plugin uses in the admin is now a proper enqueued file rather than a tag written into the page. Nothing changes on screen; it means the scripts are versioned, cached, and can be removed by a site that needs to.
- Eleven more settings can now be changed from WordPress instead of only from your account area: the player's custom CSS, an audio jingle before the article, background music and its volume, an advertising break inside long items, a different jingle per category, the pronunciation lexicon, and the five texts a visitor sees when your audio is reserved.
- A default voice per language. This one belongs to your account rather than to this site, and it applies according to the language of each item.
- The pronunciation lexicon is written as plain lines, "written = pronounced". A line without an equals sign is ignored rather than guessed at, because an invented pronunciation would be heard on every article.
- Settings that exist on your account but not on this site are never erased by saving here. Jingles set for a category of another one of your sites, and voices set for a language this screen does not show, are kept.
- Fixes a setting that could be lost without anyone touching it. A player layout or a voice this screen could not display was replaced by the first item in the list, and the next save sent that instead, even if you only came to change something else. Anything the screen does not recognise is now shown as it is and left alone.
- The player layouts are now read from your account instead of being listed in the plugin. The four newest ones (guided reading, floating bar, summary first, episode) can be chosen from here, and any layout added later will appear without updating this plugin.
- Each layout says what it needs from an item before you pick it, rather than after. An item that cannot supply it is served with the compact bar instead, with no error, which used to look like a broken plugin.
- Housekeeping for the WordPress.org directory: the plugin URI and the author URI are now two different addresses, as the directory requires. The plugin is published by Tout Est Faisable, which operates the WeDispatch service.
- The plugin folder is now
wedispatch-audio-text-to-speech, matching the name and the directory address. If you installed an earlier version by hand from wedispatch.fr, WordPress sees the new folder as a separate plugin: deactivate and delete the old one after installing this one. Your settings and the link between each item and its audio are stored in the database, not in the folder, so nothing is lost.
- The plugin is now called "WeDispatch Audio Text to Speech".
- French typography: a label and its explanation are now separated the way French does it, with a space before the colon. The colon lived in the template, outside any translatable string, so no translation could fix it.
- The screenshots on this page were retaken on the current version.
- Bylines can now be added to items that already have audio, and it costs nothing: a new card on the overview posts them in one click. The text has not changed, so no audio is produced and nothing is taken off your monthly volume. This supersedes what 1.20.0 said about bylines not being retroactive: the service changed on the same day.
- The plugin now tells the service which sends are automatic and which are asked for. Only the automatic ones (an item published in Auto mode) go under the "Automatic audio" switch of your account and of the site. A click on "Generate", a bulk action or a queue from a screen always goes through, even on a site that is paused.
- An item that was not turned into audio because that switch is off now says so, on the Content screen, instead of showing an unexplained "No audio". It is also queued again correctly once the switch is back on.
- Manual mode is what stops this site from sending anything on its own, and the setting now says it.
- Listening figures by byline. A new table on the Statistics screen shows, for each byline, how many items, how many listens and how many were heard to the end. Items with no byline keep their own row, so the rows add up to the totals above them.
- The plugin now sends the byline with each item, taken from the author's display name. It sends nothing when it cannot tell, because a guessed byline enters a newsroom ranking and nobody would know it was guessed. Co-Authors Plus bylines are read when that plugin is active, and the
wedispatch_signaturefilter lets a site send anything else ("The newsroom", a pen name, a custom field). - Bylines are not retroactive: items already turned into audio stay unsigned until their next generation. Correcting a byline never costs a generation.
- You can disconnect the site. Until now you could connect but never leave, except by emptying the API key field by hand. The confirmation says exactly what goes (the key, everything read from your account, the listening secret and its hourly refresh), what stays (your settings, and the link between each item and its audio, so reconnecting puts everything back), and what it cannot do for you (the key keeps existing on wedispatch.fr until you revoke it there).
- Choosing or changing your monthly volume now goes straight to your account area, where the slider lives.
- Fixes "The link you followed has expired" when connecting the site in one click. Nothing had expired: the return address was escaped for HTML, so the security token arrived under a mangled name. One-click connection had never worked end to end; it does now.
- Fixes pasting an API key by hand. The plan, the entitlements and the usage of the previous account stayed on screen for five minutes, which read as "my new key is not being taken". Changing the key now forgets everything that described the old account.
- You can now hear a voice before choosing it. Every voice list has a Listen button, wherever it appears: our screens, the Categories screen, an author's profile, the editor panel. Picking a voice from a name and two adjectives was guesswork.
- "My subscription" shows the volume you actually subscribed to, in items per month, and so does the status badge on every screen.
- A new card on the overview: what is left this month, which published items have no audio yet, and one button to queue them. Unused items are lost at the end of the month, so this uses what was about to disappear. It never queues more than you have left.
- The page cache is emptied when an audio is attached. "I generated the audio and the page does not show it" is almost always a cache serving the version from before. WP Rocket, LiteSpeed, W3 Total Cache, WP Super Cache, Cache Enabler and Cloudflare are handled; the Help screen names the one it found, or says it found none.
- Subtitles download as SRT as well as VTT.
- One offer, and one question: how many items per month. There is no longer a tier to compare, so every feature is included whatever your volume. "My subscription" now counts in items rather than characters, shows how many are left, and says plainly that unused items are not carried over to the next month.
- The WeDispatch screens now carry the product's own colours, the same ones as your account area. Settings on this site and settings on wedispatch.fr no longer look like two different products. Only our own pages change: your admin menu, your toolbar and every other plugin are untouched.
- A voice per category is part of a paid subscription, and the screens now say so before you set one rather than after. Every category voice already set keeps working and stays visible: nothing is erased, and it all comes back the day the subscription does.
- A voice per author is unchanged and stays open to everyone.
- Connecting your site no longer means copying an API key by hand. One button, one authorisation page on wedispatch.fr, and the site is connected. No password is ever typed into WordPress, and the key never travels through a browser.
- Works on staging, on an intranet and behind HTTP authentication: your site does the calling, so nothing has to reach it from outside.
- The default speed list no longer offers two choices that do the same thing. An unset speed and "Normal" both play at exactly 1: you picked one, heard no difference, and started doubting the rest. An account that never set a speed now simply shows "Normal", which is what it already sounds like.
- The font list said "Inherited from the player" on a screen called Player, which explains nothing. It now says what it does: do not impose a font. That option is not a duplicate of "System"; one leaves the player its own font, the other imposes the system stack.
- Fixes a "Contact support" link that led to a page which had never existed, on the Help screen, which is exactly where someone with a problem goes. It now opens an email to the support address the rest of the site uses.
- The Player screen now says "this screen is ahead of the service" instead of showing a bare 404 when the WeDispatch service has not been updated yet. The plugin and the service are released separately, so that window always exists. Nothing is broken in it: the player works and its settings are untouched, and the screen starts working on its own.
- A route that does not exist yet and a service that is down are opposite pieces of news: one asks you to wait, the other to act. They no longer look the same.
- The connection state is now proportional to the news: a quiet pill in the header when everything works, a full width banner the moment something does not. Six green banners a day train people to ignore the red one.
- Screens no longer stretch across a whole wide monitor: a table that runs 2500 pixels loses the eye between the title on the left and the date on the right.
- My content & audio: the voice source is shown only when it is an exception. On a site with one voice it used to repeat "site voice" on every single row, which hid the two rows where it actually differed.
- New screen: Player. Layout, font, default speed, highlighting, summary, download, sharing buttons. Until now these lived only in your WeDispatch account, and this plugin could do nothing but tell you to go there.
- A setting your plan does not include is shown disabled and explained, never offered and then refused when you save. A form that proposes what it cannot deliver is the fastest way to make people doubt everything else.
- "Who can listen" is shown here but changed on your account, on purpose: switching it moves every audio file you have ever published from one storage area to another, and done halfway it would silence your whole archive.
- Listening samples: where the service provides one, voices can be heard before they are chosen.
- Transcript. Optional, folded, under the article. It is what makes the audio reach a deaf or hard of hearing reader, which a player alone never does, and it is text that search engines index. Built from the word-level timing already captured for highlighting: no AI, no extra cost. Downloadable as .vtt or .srt from My content & audio.
- "Listen next". Optional, under the article: same category first, then the most recent. The list depends on the article being read, never on who is reading it, and only content published on this site is suggested.
- Both are off by default. Adding things under someone's articles without being asked is not an improvement.
- Bulk generation now sends up to 25 items in a single request instead of one background task per item. On a low traffic site a batch of fifty used to drain at the speed of visitors.
- Statistics: content produced from another site with the same key is now listed separately, below your own. On a shared key it used to fill the whole table, and nobody could find their own articles.
- Statistics: fixes a total that was smaller than its own 30 day period. Two different definitions of the word "listen" had been coexisting on the service side.
- New screen: Statistics. Listens over the last 30 days against the 30 before, listens since the beginning, and how many people listen to the end.
- Your most listened content, named the way you name it: the figures are matched back to your posts and pages, so each line is a title you recognise with a link to edit it, not an internal identifier.
- A figure that is structurally zero is not shown. A site that reserves nothing for subscribers does not need to read "subscribe prompt conversion: 0%", which is exactly true and reads like a fault.
- Content counted by the service but absent from this site is still listed, and said to be so: deleted here, or produced from another site using the same key. Hiding it would leave a total that never adds up.
- Figures are cached for 15 minutes, with a Refresh button, because counting them costs the service real work.
- New screen: Voices. Who reads what, on one page. The site voice, a voice per category and a voice per author, all set and saved together.
- Until now those three settings lived in three unrelated WordPress screens: ours, the Categories screen, and each author's profile. Almost nobody found the last two, so the feature that sets this plugin apart went unused.
- The screen opens on the routing itself: the four steps WeDispatch goes down to pick a voice, and how many rules you have set at each one.
- Only the people who have actually published in the content types you read aloud are listed, so a shop with ten thousand customer accounts still gets a short list.
- If the voice catalogue cannot be read, the page shows your rules but refuses to let you save. An empty list of voices would quietly erase every choice already made.
- The Overview screen now agrees in number: "1 item with audio", not "1 items with audio".
- When the site is not connected, Overview no longer ends on empty space. It shows the next step, worded for the situation at hand: no key yet, a key that is refused, or a service that could not be reached.
- New screen: My content & audio. Every post, page or product on the site, with its audio state, the voice that will read it, and what to do next. Generate or rebuild one item, or several at once. Filter by content type, by audio state, or search the titles.
- The list is built entirely from WordPress, with no network call. The service is only asked to do work, never to describe your site.
- Bulk actions are queued in the background rather than run inside the page, so a batch of fifty does not time out halfway with nobody knowing what went out. The screen says "queued", because that is what happened.
- An item whose text changed since its audio was made is shown as such. Its audio keeps being served: that is a decision to take, not a failure to repair.
- Fixes a stylesheet that failed to load in 1.11.0, which left the new screens unstyled.
- WeDispatch now has its own place in the WordPress menu, instead of hiding under Settings. Four screens to start with: Overview, Settings, My plan, and Help & checks.
- The Overview screen answers the three questions you actually have: is it connected, what is it costing this month, and what is left to set up. Every item in the getting-started list is computed from this site, not copied from a manual.
- The connection state is now at the top of every screen, and it tells the difference between a refused key and an unreachable service. Those two need opposite reactions, and confusing them sends people to regenerate a key that was never the problem.
- Help & checks: a one-click key check, plus a diagnostics table you can paste into a support request.
- The old address (Settings, then WeDispatch) still works: it redirects.
- The settings screen now names the WeDispatch address this site talks to, right under the API key field. A key created for a different WeDispatch environment is refused with "invalid key", which is true and tells you nothing about the cause.
- Posts are no longer the only thing that can be read aloud: pages, products and any custom post type can be, each one ticked in the settings.
- New WeDispatch audio player block for the editor, and a widget of the same name for sidebars, alongside the shortcode, which is unchanged.
- Every screen of the plugin is now translated. A French translation ships with it, and any other language can be contributed on translate.wordpress.org.
- The player is exposed to screen readers as a labelled landmark, so it can be announced and jumped to instead of being a silent block at the top of the article.
- Requires WordPress 6.3 or later (the editor block uses the current block API).
- Settings: the reading mode selected on the settings page, Automatic or Manual, is stored and applied exactly as chosen.
- Conditional access: the player identifies a signed-in reader through the REST route on sites served by a page cache, so audio follows the same access rules as the text.
- The player script now goes through the standard WordPress script queue: it is deferred and placed in the footer, so caching and performance plugins can handle it like any other asset. Its position in the article is unchanged.
- Every setting is now validated against a closed list of accepted values before being stored, and the API key is trimmed of the stray spaces a copy-paste often carries, the most common cause of a "key refused" message.
- Form tokens added to the per-author and per-category voice fields.
- Readme rewritten in English and tested against WordPress 7.0, as required by the official plugin directory.
- Every item is read in its own language. The plugin determines it on its own: Polylang, WPML, or the site language for a single-language install. A
wedispatch_post_languagefilter lets you set it differently. - Voices are now grouped by language in the selection lists, with the language of the item first, and the account's signature voices at the top.
- Scheduled items are handled from the moment they are scheduled: the audio is prepared ahead of time and goes live at the planned hour, together with the text.
- New setting: automatic regeneration when the text of an item that already has audio changes. Off by default, since each regeneration uses one generation.
- Every request carries an idempotency key: a call replayed after a network drop produces neither a duplicate reading nor a duplicate charge.
- The "Generate audio" button now regenerates an item whose text has changed since the last audio version, instead of announcing a preparation that was not happening. The status shown in the editor panel now distinguishes audio ready, preparation in progress and scheduled go-live.
- Publishing and updating no longer hold up the save: generation runs in the background and the editor is handed back immediately.
- The voice catalogue is checked before every request. A stored voice that no longer appears in the plan's catalogue is reported in the admin area and replaced by the account's default voice, instead of failing silently.
- The category travels with the item: the first category (other than the default one) is sent to the service, which applies the per-category voice, jingle and statistics configured in your account area.
- Self-contained access lock: the plugin signs the visitor token itself (no code needed on the editorial side), with the secret retrieved from the service; a "Check that everything works" button in the settings.
- Preparation for the WordPress.org directory: external service declaration, complete headers (license, required versions), capability check on the per-category save, hardened handling of form data.
- "My subscription" panel in the settings: plan, monthly usage (bar), included rights and a suggested upgrade, with "subscribe" and "manage my subscription" shortcuts.
- Editor panel: preview of the voice actually used in Auto mode (category, author, site routing) and clearer mode labels.
- Voice per category (taking priority over the author), voice per author, mode meta box and manual generation.
- First release: generation on publication, automatic player, connection test.