Audio for French-speaking Swiss media: where our servers are, and what it changes for a Swiss publisher
Switzerland is not in the European Union. Where WeDispatch hosts data and builds the audio, what path a Romandy publisher's text takes, and what the sovereign tier keeps inside Europe. Stated without rounding.
French-speaking Switzerland is a small market with a high density of media: many titles, many demanding readers, and one constraint French publishers never have to weigh, that of hosting outside the European Union. For a Romandy title that wants to offer its articles as audio, the real question is not the quality of the voice, it is: where does my text go, and what does that imply under Swiss law. Here is our path, plainly, and what to do with it.
Where our servers are
Let us say it right away: our servers are not in Switzerland. The application data and the audio files produced are hosted in the European Union, in Ireland, with the provider Supabase. The site and API are served by Vercel. Text preparation, the stage that sees your entire article (cleanup, extraction from the page, image description), is run by Mistral AI, a company established in France: it therefore happens in Europe, without leaving the Union. For a Swiss publisher, this means the text crosses the border: it leaves Switzerland to be processed in the European Union. That point deserves to be stated clearly rather than left to guesswork, as we already detail in where your text actually goes.
The stage that goes further: synthesis
Building the voice itself, from the prepared text, is run according to your account's audio tier by Cartesia AI or by Google, companies established in the United States. On those tiers, the prepared text is passed to them to produce the audio file, which constitutes a transfer outside the European Union. We write it rather than hide it. For a Romandy publisher, the full path on the standard tiers is therefore: Switzerland to European Union for preparation, then European Union to United States for synthesis. Three regimes, one action on your part, but one you are better off knowing before you publish.
What the sovereign tier keeps in Europe
For newsrooms that cannot let a text leave Europe, there is a sovereign tier: on it, generation fails rather than switching outside the European Union. It is a deliberate choice, the opposite of the silent fallback that would quietly send the text elsewhere so that "it works anyway". On this tier the path stops at the Union: Switzerland to European Union, and nothing beyond. A Swiss publisher subject to strict requirements on where processing happens therefore has an option that keeps its promise, and above all an audio produced with, in its provenance log, the processing location actually used. Traceability, article by article, is what separates a guarantee from an intention.
What it implies, under Swiss law
We are neither your lawyer nor your data protection officer, and this article does not replace their advice. The verifiable fact we give you is the exact path of the text and the location of each stage. From there, the revised Swiss Federal Act on Data Protection (nFADP), in force since September 2023, governs the disclosure of personal data abroad, and it is for the publisher, with its advisers, to check that the path described here is compatible with its own obligations and the kind of content involved. For an already public news article, the stakes are low; for embargoed content or an internal memo, the question genuinely arises, and the sovereign tier is there to answer it. Our French text to speech page sums up the spirit of this approach: say where the text goes, do not promise in general terms that it is "secure".
One last point that often reassures
The audio player embedded in the site collects no personal data on listeners: no cookie, no IP address written down, no tracker. A listen is counted per article, and nothing else. For a Romandy title mindful of what it imposes on its readers, that is one less thing to manage on the consent banner. The data question does not stop at the text you send: it includes what the tool does, or does not do, to the people who listen. Here too, the right answer fits in one verifiable sentence, the same logic that underpins our piece on monetising audio on a news site: first say what happens, only then talk about it as an advantage.
Give your articles a voice with WeDispatch
This blog is itself voiced by WeDispatch. Curious how it sounds on your content?
Book a demo