voixta

Voixta Privacy Policy

Effective date: 15 November 2026

1. Who we are

Voixta is an online tool that turns the speech in your video into captions, including when the speaker switches language mid-sentence. It is currently a free, invite-only pilot. This policy covers both the public waitlist and the pilot service.

2. What we collect

If you join the waitlist

  • First name, last name and email address.
  • Your role: content creator, reel maker, editor or other.
  • The interface language of the page you used.
  • Campaign tags from the link you came from (utm_source, utm_medium, utm_campaign), if any.
  • When you signed up, when you gave consent, when you confirmed your email address, your status (pending, invited, unsubscribed) and when you were invited.
  • Your IP address, used for the bot check and rate limiting. We store it only as a one-way hash, and we delete those rate-limit records after 1 day.
  • If we add you to the waitlist ourselves (for example, someone who told us in person that they want to try Voixta), we record that we added you and when.

Joining the waitlist is a two-step process. After you submit the form, we send you an email with a confirmation link. We add you to our mailing list only after you open the link and press the confirm button. If you never confirm, we send no further emails except a resend of the confirmation email if you sign up again. Joining the waitlist does not give you access. We email selected people to invite them.

If you are invited to the pilot

  • Account data: email address, sign-in data, interface language, your monthly minutes allowance and the minutes you have used.
  • The videos you upload and the audio extracted from them.
  • Transcripts, captions, your edits and your caption styles.
  • Your personal word list: names, brands and other words that transcription keeps getting wrong. It contains words you add and words we learn automatically from your own caption fixes (if you make the same fix in two projects, the word is added). The list is yours alone and is never shared with other users. We send it to our transcription providers as spelling hints.
  • Videos rendered with burned-in captions.
  • Whether you are a pilot participant and any pilot benefit we have agreed with you.

When you visit the site

Aggregated usage statistics from Vercel Web Analytics: the pages you view, and a few actions on the landing page (clicking a “join” link, starting the form, signing up, or a failed signup and its reason). We keep only campaign (utm_) tags from the page address and drop everything else.

What we do not do

  • We do not sell or rent your data, and we do not use it for advertising.
  • We do not use recordings to identify people (no voice biometrics).
  • We make no automated decisions about you with legal or similarly significant effect.
  • We do not charge for the pilot and do not collect any payment data. If we introduce paid plans, we will update this policy first.

3. Cookies

We use only strictly necessary cookies and similar technology, so we do not show a consent banner:

  • sign-in session cookies;
  • a cookie that remembers your interface language;
  • technical data that Cloudflare Turnstile may set to tell humans from bots.

Our analytics (Vercel Web Analytics) works without cookies: it counts visitors using a hash of the request, not a stored identifier. We set no advertising or tracking cookies.

4. Why we use your data, and on what basis

PurposeLegal basis (GDPR)
Sending you the confirmation email, then emailing you about early access, the pilot and launchYour consent, Art. 6(1)(a), confirmed through the link in our email
Providing the service to invited users: processing your uploads, making captions, rendering and exporting, tracking your minutesContract, Art. 6(1)(b)
Keeping your personal word list and using it to improve your captionsContract, Art. 6(1)(b)
Emailing you a warning before we delete your inactive dataContract, Art. 6(1)(b)
Proving you signed up and confirmed (time of signup and confirmation)Legitimate interest, Art. 6(1)(f)
Security, bot protection, rate limiting and preventing abuseLegitimate interest, Art. 6(1)(f)
Understanding, in aggregate and without cookies, how visitors use the site and which links bring themLegitimate interest, Art. 6(1)(f)

5. Who receives your data

These service providers process data on our instructions. We have data processing terms with them, and they may use your data only as described in this policy, with the exception of Deepgram described in section 10.

ProviderWhat it receivesWhyLocation
SupabaseAccount and waitlist data, transcripts, settingsDatabase and sign-inUS company; data region Ireland (eu-west-1)
VercelEvery web request, including IP address; aggregated analytics eventsHosting the site and app; cookieless analyticsUS
Cloudflare (Turnstile)Browser signals, IP addressBot check on formsUS
Cloudflare (R2)Uploaded and rendered videosFile storageUS company; storage location hint: Eastern Europe
ResendName, email address, the email contentSending our emails (confirmation, welcome and invitations)US
DeepgramAudio track, word list hintsSpeech-to-textUS
SpeechmaticsAudio track, word list hintsSpeech-to-textUK
Microsoft Azure (Speech)Audio track, word list hintsSpeech-to-textUS company; region Sweden Central
Amazon Web Services (via Remotion Lambda)Your video, its audio and your captions while we process themScaling videos down, extracting audio for transcription, and export with burned-in captionsUS company; region Frankfurt (eu-central-1)

Which speech-to-text provider handles a video depends on the languages in it. Only the audio is sent, never the picture. Speechmatics keeps the audio and its transcript for up to 7 days and then deletes them automatically.

We disclose data to authorities only when the law requires it.

6. Transfers outside the EU/EEA

Most of the providers above are in the United States, and Speechmatics is in the United Kingdom, although we use its EU1 region for processing. Cloudflare, Vercel, Microsoft, Amazon (for AWS) and Resend are certified under the EU–US Data Privacy Framework, so for transfers to them we rely on the European Commission’s adequacy decision. Supabase (which stores our data in Ireland) is not certified, so for it we rely on the EU Standard Contractual Clauses in its data processing addendum. Deepgram is not certified, so for it we rely on its standard terms and its own transfer safeguards. For the UK we rely on the European Commission’s adequacy decision.

7. How long we keep it

DataRetention
Unconfirmed signups (you never clicked the link)60 days after signup, then deleted
Confirmed waitlist entryUntil the pilot ends or you unsubscribe, whichever comes first. If you unsubscribe, we delete your other details and keep only your email address for 1 year, so that we do not contact you again.
IP hashes used for rate limiting1 day
Proof of your consent (signup and confirmation times)3 years after it is no longer needed for emailing you, to defend legal claims.
Aggregated analyticsUp to 12 months
Accounts, uploaded videos, transcripts, word lists30 days after you last use Voixta. We email you 7 days before deleting, and we can extend the period on request.
Rendered videos30 days after the render finishes. We can extend the period on request.

You can delete your account at any time, and we then delete your data within 14 days, except what we must keep by law.

8. Your rights

You can ask us to give you access to your data, correct it, delete it, restrict our use of it, or object to our use of it. You can also ask for a copy in a portable format.

You can withdraw consent at any time through the unsubscribe link in every email or by writing to support@voixta.com. This does not affect use before you withdrew.

We reply within one month. You can complain to the Lithuanian State Data Protection Inspectorate (Valstybinė duomenų apsaugos inspekcija) or to the data protection authority in your own country.

9. Recordings of other people

Upload only videos you have the right to use, and only with the consent of the people who appear or speak in them. We process recordings only to make your captions.

10. AI models and training

We do not use your recordings or captions to train our own AI models. Our speech-to-text provider Deepgram may keep and use the audio it receives to improve its own models, under its Model Improvement Program. Speechmatics does not use customer audio for training unless the customer opts in, and we have not. Microsoft states in its documentation that it does not retain or store the audio it processes for our transcription requests.

11. Children

Voixta is not intended for anyone under 16. If a child has signed up, contact us and we will delete the data.

12. Security

We encrypt data in transit and limit access to the people who need it. Database access is restricted to our server, and the public cannot read the waitlist. At present, only one person has that access.

13. Changes

We will post updates here with a new effective date and notify you of material changes.