Privacy
Privacy policy
The short version
- Dicta Dork turns your voice into text on your iPhone itself.
- You don't need an account.
- The app contains no analytics and no trackers.
- Your text only leaves your iPhone when you send, copy or share it.
- The app makes only two connections of its own: downloading the speech model, and checking for an update when you tap that button.
- This website sets no cookies, and my server keeps no record of visits.
1. Who I am
Dicta Dork is an iPhone app. Until October 2026 it was called Sotto. I make the app and this website as a private individual: Ruud Hoevenaar, the Netherlands. There is no company and no registration with the Dutch Chamber of Commerce. My Apple developer account is in my own name.
You can reach me at [email protected]. To report something about the app, mail [email protected] or [email protected]. All three end up in the same inbox.
This policy exists in English and in Dutch. Both versions say the same thing.
2. What stays on your iPhone
This data is stored only on your iPhone, in the app's private storage. I never receive it.
- Your recordings, including the ones you start with Siri.
- The text of your dictations and your history.
- Your word list.
- Your settings, such as the recipient and subject of your email, the chosen model and your retention periods.
- The speech model you downloaded.
Speech is turned into text by a speech model on your iPhone, using Whisper through whisper.cpp. Your recording is not sent to a server for this, and not to anyone else's AI service.
3. How long the app keeps your data
You choose this yourself in Settings. These are the defaults.
- Your recordings: 7 days after transcription. You can also choose: until the text is done, 30 days, or until you delete them. The app keeps the recording so you can transcribe it again, for example with the accurate model or after adding a word to your word list.
- The text of your dictations: 3 months. You can also choose 1, 6 or 12 months, or keep it forever.
- Your word list, your settings and the speech model: until you change or delete them.
If a transcription fails, the recording stays until you transcribe it again or throw it away.
4. When the app goes online
The app makes only two connections to the internet of its own.
- Downloading the speech model. The app asks first and shows the size. The model comes from Hugging Face (Hugging Face, Inc., United States), through
huggingface.coand their delivery network. Hugging Face sees your IP address and the technical details of the request, such as the iOS version in the User-Agent. The app sends no audio, no text, no name and no email address. I don't receive this data. - Checking for an update. Only when you tap ‘Check for update’ in Settings. The app then asks Expo's update service (650 Industries, Inc., United States) at
u.expo.devwhether there is a new version. What goes along: your IP address, the platform (iOS), the app version, the build's channel, and a random number that belongs to your installation. That number is stored on your iPhone and stays the same while the app is installed. It is not derived from your device identifier or from an account. The app sends no audio, no text, no name and no email address. This feature only works in the version from TestFlight or the App Store.
The app never checks for updates on its own. Apart from these two, the app does not connect to the internet: there is no other network code in the app, and no library that sends anything in the background. Once the model is downloaded, the app works without internet.
See also Expo's privacy policy and Hugging Face's privacy policy.
5. When you send, copy or share
Your text only leaves your iPhone when you choose that. Dicta Dork never sends anything by itself.
- Sending through Outlook. Dicta Dork opens the Outlook app on your iPhone with a draft: recipient, subject and text. You do the sending, in Outlook. Dicta Dork can't see whether that worked. From that moment on, the terms of Microsoft and of your email account apply. Outlook may, for example, save a draft on your email provider's server.
- Copying. The text goes to the iOS clipboard. If you use Handoff, iOS can share the clipboard with your own other Apple devices. If your text is longer than 4,000 characters, ‘Send through Outlook’ also puts it on the clipboard, because Outlook may cut off a long text.
- Sharing. In the iOS share sheet you pick the app or the person. That app decides what happens to your text next.
- Report a problem. In Settings › Help, ‘Report a problem’ prepares an email to [email protected] (beta 0.3.0 still uses [email protected]; that address keeps working), in Outlook or otherwise in your iPhone's mail app. The subject contains the app version and build number. The bottom of the mail repeats those, plus your iOS version and your iPhone model, such as ‘iPhone 15 Pro’. Dicta Dork adds no text from your dictations, no audio and no attachment. You see the mail before you send it, and you can change or remove anything. You do the sending, from your own email account. Nothing goes out by itself. How the mail reaches me and how long I keep it is in section 11.
6. Microphone and Siri
The app asks for microphone access the first time you record. The microphone is only on while a recording is running. You can withdraw access at any time: Settings → Dicta Dork → Microphone.
A recording starts in one of two ways: you tap the button, or you tell Siri. Siri can start a recording without you opening Dicta Dork. If iOS allows it, that also works on a locked iPhone. That recording stays on your iPhone until the app transcribes it. A recording stops after 30 minutes, when you stop it, or on an interruption such as a phone call.
If you say ‘Start Dicta Dork’ or ‘Stop Dicta Dork’ (beta 0.3.0 still uses ‘Start Sotto’ and ‘Stop Sotto’), Apple processes that Siri request under Apple's privacy policy. The app only receives the command. Dicta Dork records the dictation itself, not Siri.
7. Analytics and trackers
The app contains no analytics, no usage statistics, no ads, no trackers and no crash reporting by me or by anyone else. There is not a single telemetry library in the app.
The app does not ask for access to your contacts, your location, your photos or your calendar.
8. Your iPhone backup
When you back up your iPhone, with iCloud or a computer, your settings and your word list are included. That backup falls under your own Apple account.
The app excludes your recordings, the text of your dictations and the speech models from the backup. So they are not in it. If you restore the backup to a new iPhone, you get your settings and word list back, but not your recordings, not your texts and not the model. You download the model again.
9. Beta through TestFlight
While Dicta Dork is in beta, testing runs through Apple's TestFlight. Apple then shares data with me through App Store Connect:
- Your name and email address, if I invited you by email.
- The date you installed the app, how often you opened it and the number of crashes.
- The feedback and screenshots you send yourself.
- Crash reports, if you allowed that in iOS: Settings → Privacy & Security → Analytics & Improvements → ‘Share With App Developers’. These include technical details, such as your iPhone model and iOS version.
None of this contains text from your recordings, unless you put it in your feedback yourself. Apple does receive the name you gave your device, but does not pass it on to me. I use this data only to improve the app and share it with no one. I keep it for as long as the beta runs. After that I remove the testers from App Store Connect.
10. This website
This website uses no cookies, no analytics, no trackers and no files from other domains. There is no JavaScript on the site. The font is served from the server itself.
I host the site myself, on my own server at Contabo GmbH in Germany. Traffic passes through Cloudflare (Cloudflare, Inc., United States), which delivers the site and protects it against attacks. Cloudflare sees your IP address and the technical details of your request, and processes them under Cloudflare's privacy policy. I have not switched on any Cloudflare measuring scripts; in Cloudflare's dashboard I only see totals, such as the number of requests per day.
On dictadork.com, Cloudflare’s Network Error Logging is off: Cloudflare does not ask your browser to report network errors. On the old address sotto-dictate.com, which only forwards you here, it may still do that. That is only about failures, not about normal visits.
My server keeps no record of visits. Request logging is off, on the web server and on the reverse proxy in front of it. Your IP address is needed to send the page to you — that's how the internet works — but it is not written down anywhere, so it is not kept either. There are no visitor records, no statistics and nothing to look back through. Only real malfunctions end up in a technical error log; that is capped at two files of 1 MB, which overwrite each other.
11. When you email me
The addresses feedback@, support@ and [email protected], and the same addresses on the old domain sotto-dictate.com, run through Cloudflare's email service (Cloudflare, Inc.) to my own inbox. Cloudflare forwards the mail and may hold it briefly on the way.
If you email me, I receive your email address, your message and whatever you attach. An email sent through ‘Report a problem’ also contains the app version, your iOS version and your iPhone model (see section 5). I use that only to answer you and to improve the app. I keep your mail for 12 months after your question is resolved. Please don't include confidential text that your question doesn't need.
12. Deleting your data
In the app you do this yourself:
- One dictation: open it in History, tap ‘More’ (…) and choose ‘Delete’.
- All dictations: History → ‘Clear all’.
- Only the stored audio: Settings → Keep audio → ‘Delete all audio now’. Your texts stay.
- A speech model you don't use: Settings → Speech model → the bin icon next to that model. The model in use disappears when you delete the app.
- Everything: delete the app. iOS then deletes all your recordings, texts, settings, your word list and the model.
All I have myself is your emails and your TestFlight data. Ask me to delete them and I will.
13. Your rights
You have rights under the General Data Protection Regulation (GDPR): access, rectification, erasure, restriction of processing, objection and data portability. Because the app sends no data to me and this website records nothing, in practice those rights only concern the emails you send me and your TestFlight data.
To use them, mail [email protected]. I answer within a month. If you're not satisfied, you can file a complaint with the Dutch Data Protection Authority, the Autoriteit Persoonsgegevens, or with the authority in your own EU country.
14. Children
The app is not specifically meant for children and collects data from no one.
15. Changes
If the app or this website changes in a way that makes this policy incorrect, I update the policy before that version comes out. The new version appears on this page, with a new version number and a new date.
Changes per version:
- 1.2 · 28 October 2026 · The app is now called Dicta Dork (formerly Sotto). Website and email addresses moved to dictadork.com; the old addresses on sotto-dictate.com keep working (sections 1, 5, 6 and 11). The website runs through Cloudflare; section 10 still said nobody sat in between, which had not been true since 25 October 2026. The policy is now also available in English; the Dutch version is equivalent. What the app does with your data has not changed.
- 1.1 · 3 October 2026 · Described ‘Report a problem’ (sections 5 and 11). Described Siri on a locked iPhone as a possibility (section 6). Clarified deleting a dictation and a speech model (section 12).
- 1.0 · 2 October 2026 · First version.