Everything is included. There is nothing to unlock later.
Only one boundary exists: the free account on one side, the paying account on the other. Past the first euro, all 29 features are yours, whatever your volume. The only separately billed line is the signature voice, because it requires a recording session.
They are grouped by what you are trying to do, not by technical module. Each card says what it does and the problem it solves. 5 of them are marked «exclusive»: we found them at none of the competitors we surveyed, and the survey says which ones and on what date.
Reading French the way a journalist would
The heart of the product: what is read on screen is not what is said out loud. The reader sees the journalist's article, the engine receives a normalised reading, and the timestamps are stitched back so the highlighting stays right.
Your town names, officials and brands, pronounced the way you say them. The newsroom fixes it once, the product remembers.
What it solves: No rule and no model guesses Ploërmel, Cysoing or Guebwiller on its own.
If this section or this language, then this voice. The first matching rule wins.
What it solves: Picking a voice article by article stops working past ten publications a day.
The voice of one of your journalists, cloned in studio quality with their written consent.
What it solves: A named column loses its point when a generic voice reads it.
23 reading languages, from French to Japanese. One edition per language, each with its own page and declared language.
What it solves: A border newsroom or a museum cannot mix two languages in one feed.
No language model touched the text, and you can prove it. Born from newsroom AI charters.
What it solves: An AI charter usually forbids automatic rewriting, and you need to be able to prove it.
Producing without changing your habits
Your journalists publish, the audio follows. Three integration paths, a player in one line or composed by your own server, and the option of driving it all from an assistant.
22 templates that pick up your site's colours. One line of script, automatic mode by URL.
What it solves: A player that does not look like the site is a player nobody presses.
Eight admin screens, a Gutenberg block, bulk actions, and a connection with no key to copy.
What it solves: Most newsrooms will never touch an API and have no developer on hand.
For a custom CMS, with a synchronous mode that publishes the article already voiced.
What it solves: An article published then voiced thirty seconds later loses its first readers.
Your platform calls one address, receives a template and its data, and composes the player itself. No iframe, no third-party script imposed.
What it solves: Some publishing platforms forbid iframes by written rule, which made our player structurally ineligible.
Playback logic with no styling: a web component, an adapter for your existing player, iOS and Android references. You keep your design.
What it solves: A publisher with its own design system does not want a player that imposes another one.
Automatic discovery of your publications. No development, no access to your server.
What it solves: Some proprietary CMSs simply cannot be modified.
The audio is ready before it goes live, at the time you set.
What it solves: An embargo or a morning slot means the audio has to exist before publication.
The same reach as the API, relayed to an assistant: what a developer can do, it can do too.
What it solves: Newsrooms are starting to work with assistants, and the audio has to keep up.
Several titles from a single account, each with its own domain and configuration.
What it solves: A press group does not want one subscription per title.
Editing and reusing without starting over
Produced audio is not set in stone. An edit is billed like a production, one credit per thousand characters started, and the stitch lands in a silence, so you cannot hear it.
Rewrite the faulty passage: only that passage is resynthesised and stitched back in.
What it solves: A mangled name used to mean regenerating the whole article, and paying a credit for it.
A short clip taken from existing audio, ready for social, with no new synthesis.
What it solves: Sharing a seven-minute article on social media does not work.
Every spoken word is findable, with a result timestamped to the second.
What it solves: Finding one statement inside an audio archive used to be impossible.
A link that opens the audio at exactly the right moment.
What it solves: A forty-minute council report is not listened to from end to end.
Subtitle files aligned on the actual reading, not on the source text.
What it solves: Subtitles keyed to the written text drift the moment a number is spelled out.
Alt text and an audio description of an infographic, for blind readers.
What it solves: An infographic with no description is invisible to a blind reader.
Reaching beyond the article page
Your audio reaches an audience that does not come back to your site every day. Each channel is configured separately and reports its own figures.
A feed listening apps pick up on their own, with the sections you choose.
What it solves: Your readers listen in the car, and they do not open your site for that.
The listener hears an excerpt, then is offered a subscription. No cookie, no profile.
What it solves: Audio brought in no subscribers, even though it holds attention longer than an article.
Ads before and during playback, with their counts, on your own inventory.
What it solves: Monetising audio used to mean going through an outside ad network.
A discreet sound bed under the reading, adjustable per section.
What it solves: A bare voice for seven minutes tires the ear faster than a dressed-up audio.
A different voice, sound bed and rules depending on the section of the site.
What it solves: Sport and obituaries are not read the same way.
Steering the work and answering for it
What you need to work as a team and to prove your compliance, without ever tracking a listener.
Starts, completion milestones, performance by voice and by section. Anonymous counts.
What it solves: With no measurement, nobody in the newsroom knows whether the audio is worth anything.
Audio waits for human approval before it is served.
What it solves: On a sensitive story, publishing a reading nobody listened to is not an option.
Who reaches the account, with which rights, on which sites.
What it solves: A shared account ends up as a password pasted into a chat thread.
A summary of what is actually switched on, and a declaration template to publish.
What it solves: The European Accessibility Act is in force, and you need to show where you stand.
What we do not do
A list of 29 features with no stated limit helps nobody decide. Here is where we stop, deliberately, and why.
No automatic summary replaces your text, no model fixes your style. The spoken summary is an option you switch on, and charter mode proves the text was left alone.
The subscription covers audio for your published texts. Neighbouring formats, such as short cuts for social media, are handled as a supported project rather than self-service. Talk to us if the need is a recurring one.
No IP address kept, no cookie, no profile. The trade-off is deliberate: our statistics are coarser than an ad tool's, and they will not reconcile with yours.
We make the audio component compliant and we supply a declaration template. The compliance of your whole site remains yours, and our template is not legal advice.
Hosting is European, but the full end-to-end chain is still being built. We will announce it as available on the day it actually is.
A quiet month does not fill the next one, and a started block is spent. It is the least pleasant part of our model, and it is written on the Pricing page.
You will not have to choose which ones
The free account gives you enough to hear the reading on your own texts. From the first euro paid, all 29 features are open, with no tier and no negotiation.
Comparing several solutions? The seven questions to ask · How the subscription works