Glossary · 25 terms

The words of audio, defined without jargon.

For each term: what it means, why it matters when you voice articles, and the usual trap. Written for a newsroom, not for an engineer.

Technical · 8 terms

What happens between your text and the sound. That is where the quality of a French reading is decided, not in the choice of voice.

Text-to-speech (TTS)

Synthèse vocale

Text-to-speech converts written text into spoken audio. Recent neural models produce voices that are hard to tell apart from human speech.

Why it matters: It is the brick that produces the sound, but it is not where the quality of a French reading is decided. Good engines are plentiful; what separates the results is what you feed them.

The trapJudging a tool on its engine rather than on its text preparation. A gorgeous voice that reads a date wrong is still unusable in production.

Neural voice

Voix neuronale

A neural voice is generated by a neural network trained on human speech. Its natural prosody makes it hard to distinguish from a real narrator.

Why it matters: It is what makes a reading sound human or mechanical, at equal timbre quality. Over a four-thousand-character article, prosody decides whether anyone reaches the end.

The trapIt depends on the punctuation of the source text. An article with no commas and fifty-word sentences will sound breathless whatever the engine.

Word-level timestamps

Horodatage mot à mot

Word-level timestamps map every word of the text to its exact position in the audio. They power highlighting, quotes and chapters.

Why it matters: It is what makes highlighting, accurate subtitles, shareable clips and search-inside-speech possible. One piece of data, four features.

The trapThe timestamps refer to the SPOKEN text, which is not the displayed text. Using them as-is makes everything built on them drift.

Text normalization

Normalisation du texte

The step that turns written text into speakable text: numbers become words, acronyms are expanded or spelled out, abbreviations are written in full.

Why it matters: This is the real work, and it is done language by language. French has its own rules, and that is where engines go wrong, not on the timbre of the voice.

The trapAssuming a language model handles it correctly. A model that guesses can read the same sentence two different ways from one day to the next, with no way to correct it.

Prosody

Prosodie

Everything in speech that is not the words themselves: rhythm, pauses, melody, intensity. It is what makes a reading sound human.

Why it matters: At equal timbre, prosody decides whether four thousand characters are heard to the end or abandoned at the third sentence.

The trapA two-sentence demo says nothing about prosody. Monotony appears with length: test on a whole article, not on a chosen excerpt.

Phoneme

Phonème

The smallest sound unit that changes meaning in a language. It is the alphabet used to write a pronunciation when spelling is not enough.

Why it matters: It is what lets you fix a mispronounced town name ONCE, instead of respelling it phonetically in every article.

The trapRespelling a name phonetically inside the article fixes the sound and damages the text: it ships that way into subtitles, transcript and index. The fix belongs in the lexicon, not in the article.

Initialism and acronym

Sigle et acronyme

An initialism is spelled out letter by letter (BBC), an acronym is read as a word (NATO). Nothing in the writing tells you which is which.

Why it matters: French news copy is full of them, and getting it wrong is immediately audible: one botched acronym discredits the whole reading.

The trapExpanding an acronym at every occurrence is as bad as mangling it. You expand on first mention if the readership does not know it, and never afterwards.

MCP server

Serveur MCP

A server exposing a product's functions as tools a conversational assistant can call, with the rights of your own key.

Why it matters: It is what lets you say 'voice the last three sports articles' instead of opening a screen and clicking.

The trap'MCP compatible' does not say HOW MANY operations are exposed. Ask for the table: a server covering half the API will leave the assistant failing on the very action you wanted to automate.

Product · 4 terms

What your readers see and hear on the article page.

Read-along audio

Texte synchronisé

Read-along audio highlights each word of the article as the voice speaks it, letting the user read and listen simultaneously on the page.

Why it matters: It is a recognised reading aid for dyslexic readers and language learners, and it is also what holds a distracted listener.

The trapThe highlighting is often keyed to the source text, while the engine received a longer one. Without stitching the timestamps back, it drifts at the first spelled-out number.

Automatic subtitles

Sous-titres automatiques

Automatic subtitles are SRT or VTT files generated from the audio, displaying the right text at the right time in videos and players.

Why it matters: They make the audio accessible to deaf and hard-of-hearing readers, and they double as search material: a spoken word becomes a findable word.

The trapSubtitles keyed to the written text rather than to the actual reading drift as soon as a number or acronym is spelled out. They have to come from the timestamps, not the source.

Signature voice

Voix signature

A signature voice is the cloned voice of a journalist or newsroom, used with written consent to narrate the publication's articles.

Why it matters: For a named column, the voice is part of the editorial signature. A generic voice removes precisely what makes it recognisable.

The trapConsent must be written, explicit and revocable. A voice cloned without a formal agreement is a legal risk, and an internal one that costs more still.

Embedded audio player

Lecteur audio intégré

The embedded audio player is the widget on the article page that plays the audio version. It installs with one line of code or a CMS plugin.

Why it matters: It is the only part your readers see. Its look, its keyboard operation and its screen-reader compatibility decide whether it is actually used.

The trapA player that does not look like the site is a player nobody presses. Check that it picks up your colours before signing, not after.

Distribution · 6 terms

What leaves your site: listening apps, acquisition, monetisation.

Automatic podcast

Podcast automatique

An automatic podcast turns published articles into audio episodes on Spotify and Apple Podcasts, with no manual production work required.

Why it matters: It is the channel that reaches people who listen while driving or walking, and who do not come back to your site every day.

The trapA feed carrying every article becomes unreadable. Two or three well-chosen sections build a listening habit; forty episodes a day destroy it.

Audio pre-roll

Pré-roll audio

An audio pre-roll is a short ad message played before an audio article starts. It creates a brand-new advertising inventory for publishers.

Why it matters: It is the best-valued slot in audio, because everyone hears it: nobody skips a pre-roll they did not see coming.

The trapIt is also the one that makes people leave. A thirty-second pre-roll in front of a two-minute article is paid for in lost listens, and the completion rate says so before the revenue does.

Midroll

Midroll

A midroll is an advertising message inserted partway through a listen, ideally at a paragraph boundary so it never interrupts a sentence.

Why it matters: It runs once the listener is already engaged, so it drives fewer people away than a pre-roll, at comparable commercial value.

The trapPlacing it by the clock rather than by the sense. A break in the middle of a quote sounds like a fault; it has to land on a paragraph boundary.

Audio registration wall

Mur d'inscription audio

An audio registration wall plays a free preview of the audio article, then asks the listener to create an account to keep listening.

Why it matters: Audio holds attention longer than a skimmed article, which makes it a good conversion point.

The trapCounting listens per person requires an identifier. A privacy-respecting wall keeps its counter in the browser, which makes it less precise, and that has to be owned.

Audio RSS feed

Flux RSS audio

An audio RSS feed is a structured file listing audio episodes. It serves both as an ingestion channel for articles and as a podcast distribution channel.

Why it matters: It is the integration path that needs no development: you give an address, the publications are discovered.

The trapDiscovery follows what your feed exposes. A feed truncated at two hundred characters will never produce audio of the full article.

Audio monetisation

Monétisation audio

Audio monetisation covers the revenue from audio articles: advertising (pre-roll, midroll), registration wall conversion and podcast distribution.

Why it matters: Audio opens inventory you were not selling, on an audience that stays longer than on a skimmed page.

The trapMonetisation that requires a listener identifier drags you into a data regime you had avoided. That trade-off is chosen up front, not later.

Measurement · 2 terms

The figures that say whether the audio is useful, and the ones that merely look good.

Listen-through rate

Taux d'écoute

The listen-through rate measures how much of an audio article is actually heard. It is the reference metric for real audio engagement.

Why it matters: It is the only figure that says whether the audio is worth anything. Without it, the audio version stays a ticked box nobody defends at budget time.

The trapConfusing a start with a listen. A start is a click; what matters is the completion milestone, and the gap between the two is often brutal.

Audio SEO

SEO audio

Audio SEO covers what audio does for search rankings: stronger engagement and structured data, but no indexing of the sound itself.

Why it matters: A page carrying audio, a transcript and structured data is better understood, and it appears in surfaces plain text does not reach.

The trapPublishing a transcript that is not the one actually read. Two diverging versions of the same content on one page penalise each other.

Compliance · 4 terms

The obligations, and what a component can or cannot do for you.

News website accessibility

Accessibilité numérique de la presse

Digital accessibility makes a news site usable by everyone, disabilities included. In Europe, the European Accessibility Act has applied since June 2025.

Why it matters: It is no longer a public-sector topic: the European Accessibility Act is in force, and it covers private publishers too.

The trapNo single component makes a site compliant. A supplier who promises compliance by dropping in a script exposes you more than it protects you.

France's accessibility framework: the national translation of WCAG into verifiable criteria, with a mandatory public declaration.

Why it matters: It is the document a public buyer will assess you against, and it does not ask whether your site is accessible: it asks for your declaration, its date and its score.

The trapNo component added to a page makes a site RGAA-compliant. A vendor promising compliance by dropping in a script exposes you: you sign the declaration, not them.

WCAG

WCAG

The international web accessibility guidelines. Four principles, three conformance levels; AA is the one regulations adopt.

Why it matters: It is the source everything else derives from. The French RGAA and the European act invent nothing: they make WCAG enforceable.

The trap'WCAG compliant' with no level or version means nothing. Ask for 'WCAG 2.1 level AA', and a date: a 2022 audit says nothing about today's site.

European Accessibility Act

Acte européen d'accessibilité

Directive 2019/882, in force since June 2025, extends accessibility requirements to many private digital services, online news among them.

Why it matters: It is the text that turned accessibility from voluntary good practice into a requirement, for private publishers previously outside the scope.

The trapThe micro-enterprise exemption reassures wrongly: it is measured on both headcount and turnover, and does not cover a mid-sized regional title.

Commercial · 1 terms

What you are buying, and how it is counted.

Credit

Crédit

The billing unit. One credit covers an article up to a character ceiling; beyond it, one more credit per started block.

Why it matters: It is the only thing to understand in order to estimate your budget: count your articles, measure their average length, and you have your volume. Everything else follows.

The trapForgetting that a started block is spent in full. An article one character over the ceiling costs two credits, and the remainder does not roll over.

The most important term is not in this glossary.

It is the name of your town, of your mayor, or of your football club. No definition will say how it is pronounced: it is your newsroom that teaches the product, once, in the pronunciation lexicon.

This glossary is not exhaustive: it grows with the questions we are asked. Tell us the word you were missing, at contact@wedispatch.fr, and it will go in.