How we handle your audio.
Your catalogue is the whole product. This page says plainly what we do with it, who can reach it, and how to make it go away. It is written to be read rather than agreed to.
Last updated 8 September 2026
What we take in
The published audio in your feed, the metadata that comes with it — titles, descriptions, publication dates — and anything you send us directly, such as unpublished recordings or corrections to speaker names.
From that we produce transcripts, speaker labels and an index. Those derived files are what the product actually queries.
Indexing before we speak
When you give us a feed address we index what is published at it, before any conversation. We do this so the first thing you see is your own show rather than a demo. If you would prefer we didn't, write to us and we won't.
Who can see it
Your archive is yours. It isn't pooled with other shows, it isn't searchable by other customers, and it is not used to train models for anyone else. Network accounts connect only the shows on that account.
Our own staff access it only to run and repair the service, and only when there is a reason to.
Deletion
If you decide it isn't for you, we delete it: the audio we hold, the transcripts, the index and the derived files. Ask by email and it goes, including from backups as those rotate.
People in the recordings
Transcripts name guests and hosts, because attribution is the point. A person who appears in your archive may ask what we hold about them, and may ask for it to be corrected. Send those requests to us and we will handle them with you, since the recordings are yours.
The public political corpus
Verbator also indexes publicly published Swedish political audio and text — podcast feeds, SVT Play, Sveriges Radio, YouTube, and the parties' own published documents — to build the record behind Politikerna.
This corpus is separate from Studio customer archives, which are never pooled into it.
A host whose public feed is in it can ask what is held and ask for corrections or removal at [email protected], and a named person can ask what is held about them.
Nothing from it is sold or used to train models for anyone else.
Where it is kept, and by whom
Audio and transcripts sit on hardware we own and operate in Sweden, on a private network with no public ingress. Access is limited to the people who run the service. This website is served from a content delivery network with points of presence worldwide; it holds no audio, no transcripts and no account data.
Transcription, analysis and the search index all run on that same hardware. Your audio is never sent to anyone else.
One outside service sees text. We use Google's Gemini to help extract structure from transcripts; it receives transcript text, never audio.
Two more run the website and our email. Neither receives audio or transcripts:
We keep your material while your show is a customer, or while we are talking to you about becoming one. When you ask us to delete it, we delete it — the audio, the transcripts, the index and the derived files — including from backups as those rotate.
Who we are, and the law that applies
Verbator is operated by Carl-Gustav AB, org.nr 559344-8276, Västra gatan 52B, 442 31 Kungälv, Sweden. Swedish law applies, and Swedish courts have jurisdiction.
If you think we have handled personal data wrongly, write to us first — we would rather fix it than have it escalated. You also have the right to complain to Integritetsskyddsmyndigheten (IMY), the Swedish data protection authority.
This site sends basic cookieless visit signals to Google Analytics before you make a choice. These signals include limited technical information and page categories; cookieless does not mean fully anonymous. Optional analytics cookies measure repeat visits, engagement and form completion. Optional advertising consent allows advertising measurement and personalization, including Google Signals where activated. We never send form contents, quiz answers or political matches to Google. You can choose each purpose separately or keep basic measurement only, and change your choice through “Analytics choices” in the footer. Your preference and cookies last up to six months. Event data is retained for two months. Google may process information outside the EU/EEA; see Google’s privacy policy. The front-page form also uses Cloudflare’s anti-spam check.
Asking us anything
One address, a person reads it: [email protected].