Quebec French and text to speech: what a France-tuned voice reads well, and what it doesn't
A Quebec French text read by a voice built for the French of France: what comes out clean, where the accent gives the tool away, and what you can and cannot fix. One Quebec paragraph put under the microscope, nothing hidden.
A text written in Montreal and a text written in Lyon look almost the same on the page. They do not sound the same. That is the first thing to grasp before you hand Quebec content to a speech tool built, like ours, for French: the question is not whether it will read the words, it will, but what a Quebec ear will recognise as its own, and what will sound like it came from Paris. Here is the straight answer, with one Quebec paragraph examined line by line.
Written French is shared, spoken French is not
Most of the gap between Quebec French and the French of France lives in speech: the accent, the prosody, the way certain vowels and consonants are realised. A speech tool, though, starts from written text. A well written Quebec article is therefore read correctly and stays perfectly intelligible: no meaning is lost. What changes is the colour of the voice. Our French voices carry the standard accent of France, not that of Quebec. Saying so upfront heads off the most common misunderstanding: the reader who expects a local accent and hears Paris instead.
What the tool reads without stumbling
Vocabulary specific to Quebec is, in itself, no obstacle. Magasiner, courriel, stationnement, fin de semaine, presentement, depanneur, cellulaire: these are French words with French spelling, and the engine pronounces them correctly with no special setup. The same holds for numbers, which in Quebec follow the France system (soixante-dix, quatre-vingt-dix), unlike the septante and nonante of Belgium or Switzerland. An everyday Quebec text, with no exotic proper nouns and no English loanwords meant to be said the English way, comes out clean and understandable on the first pass.
Where you can hear it
Take a deliberately marked sentence: "En fin de semaine, on a magasine a Chicoutimi, pis on est passes au depanneur avant de reprendre le char pour l'Outaouais." The tool reads it with no error of meaning, but three things give away that it is not from here. First, assibilation: in Quebec French, t and d before the vowels i and u are realised with a light "ts" and "dz". Our engine does not reproduce it; it pronounces those consonants the French way. Second, the vowels: Quebec French keeps and diphthongises sounds that the French of France has flattened, and our voice renders them the France way. Third, place names, Chicoutimi, Outaouais, Kamouraska, Chibougamau, are the real tipping point: there, the risk is no longer accent but an outright mispronunciation. It is the same problem as any proper noun or place name in audio, just more frequent in a Quebec text.
What can be fixed, and what cannot
The distinction is sharp, and it decides everything. Proper nouns can be fixed: our targeted correction lets you lock in the intended pronunciation of a word once, and the tool honours it everywhere after that. A Quebec publisher narrating articles sets their recurring toponyms in advance, exactly as a French regional outlet sets its own. The accent, on the other hand, cannot be set: our voices are French of France, and we do not offer a Quebec-accented voice today. If that is what you are after, a voice that "sounds local" for a local audience, the tool does not do it yet, and it is better to know that before you test than after. It is the same line our French text to speech page takes: it is tuned to read one language well, not to imitate all of its regional variants.
Our position, plainly
A Quebec text handed to WeDispatch is read correctly, stays fully intelligible, and is better for having its proper nouns set once. What it will not have is the Quebec accent: the voice stays French of France. For many uses, trade press, institutional content, professional blogs, that intelligibility is enough and the accent matters little. For an outlet that makes spoken proximity part of its pitch, it is a real limit, and we would rather write it than paper over it. That honesty about what the tool does not do is the same one we defend in what French quality means for a voice tool: we say what we read well, and we say where the tuning stops.
Give your articles a voice with WeDispatch
This blog is itself voiced by WeDispatch. Curious how it sounds on your content?
Book a demo