Privacy Policy
Last updated: 2026-10-21
WhisperType is built to keep voice dictation private by default. Our core product processes audio locally on your device, not on our servers.
1. What WhisperType does with your data
WhisperType converts speech to text on your own machine. Audio is processed locally and is not uploaded to WhisperType servers as part of normal product use.
While you dictate, the audio is written to a file on your machine, so a crash or a closed app does not lose it. The file is deleted as soon as the text is delivered. If a dictation is not transcribed, its audio stays on your machine, next to your History, so you can retry it from there: the 20 most recent, up to 500 MB, each deleted when a retry transcribes it, when you delete its row, or it passes your history retention period. This happens even with history off in Settings, so a failed dictation is never lost; delete its row in History to remove it.
1b. Optional: AI Polish (OpenAI)
AI Polish is an optional, opt-in feature that rewrites your already-transcribed text for clarity, tone, or conciseness. It is off by default. If you turn it on and choose the OpenAI provider (a local Ollama option is also available and keeps everything on-device), the text WhisperType has already transcribed locally — not your audio — is sent to OpenAI's API using an API key you supply, for that single request. This text is subject to OpenAI's own data handling and retention policies, not WhisperType's. Your audio recording is never sent anywhere, whether or not AI Polish is enabled.
2. Data we may collect
- License data. When you activate and each time the app checks license status, it sends your license key and a hashed machine identifier to our license server (trial.whispertype.ai), so a key works on one machine at a time — never your hostname or any other identifying machine data. For each subscription, the server keeps the key, the Stripe customer and subscription identifiers, the subscription status and paid-through date, and a hash of the email used at checkout so we can find your key if you lose it. A hash of your IP address is kept for two days to limit repeated requests.
- Trial signup data: the email address you provide to start your 14-day free trial (our server keeps only a hash of it), a hashed machine identifier and your app version, used only to enforce one trial per person and prevent repeat free trials. A hash of your IP address is kept for two days to limit repeated requests. Trial records are kept for about 400 days.
- Crash diagnostics, via Sentry — opt-in and off by default. If you turn this on in Settings, we collect stack traces and non-personal diagnostic tags (operating system and its version, app version, which part of the app failed). If you completed onboarding, the diagnostic tags also include the use-case tag(s) you selected during onboarding (e.g. "coding," "writing"). Audio content and transcript text are always scrubbed/excluded before any event is sent. You can opt out at any time, effective immediately.
- Usage statistics — on by default (no transcript text). To help us improve speed and reliability, the app sends performance metrics for each dictation: how long you held the key and how long the audio was, transcription latency, processing-speed ratio, model, detected language, word and character counts, success or failure and any error code, and whether AI Polish ran and with which provider. Each event also carries the app version, your operating system (Windows or macOS), a random identifier created for this installation and a session identifier. Neither identifier is linked to your email address or your license, and your IP address is not stored with the events (a hash of it is kept for up to two hours to limit abuse). Events are deleted after 180 days. This never includes your audio or your transcribed text unless you deliberately raise usage statistics to the "full" level in Settings. You can set usage statistics to "off" at any time in Settings, effective immediately.
- Basic purchase and billing records handled by Stripe, including subscription status and renewal history.
What each usage statistics level sends, for each dictation (Settings → Privacy → Usage statistics):
- Off — send nothing
- Nothing.
- Anonymous stats (no text) — the default
- Event type, a random installation identifier and session identifier, time, app version, operating system, model, language, how long you held the key and how long the audio was, latency and processing times, speed ratio, word and character counts, success or failure, outcome, status and error code, and whether AI Polish ran and with which provider. No audio, no text.
- Full — include transcripts
- Everything above, plus the transcribed text and, when AI Polish ran, the text before it was polished. Only if you pick this level. No audio.
2b. This website
whispertype.ai sets no cookies and loads no third-party scripts. The web server writes down each request: the page, the time, the site you came from, your browser, your country and your IP address. We count visitors, pages and downloads from that record. The counts hold no IP address, and the record itself is deleted after 14 days.
If you use the form “Email me the day subscriptions open”, your email address is stored on our trial server, with the date and which form you used. It is used for one email, the day subscriptions open, and nothing else. To be taken off the list, email hello@whispertype.ai.
3. Data we do not collect
- Your raw audio recordings
- Your dictated text content for server-side storage†
- Personal data sold or rented to third parties
† Two opt-in exceptions: (1) if you enable AI Polish with the OpenAI provider, your already-transcribed text is sent to OpenAI for that request — see section 1b above; (2) if you deliberately raise usage statistics to the "full" level in Settings, transcript text is included in the anonymous usage metrics described in section 2. Neither applies to the default configuration, where transcript text is never sent for server-side storage.
4. Payments and licensing
Payments are processed by Stripe. Stripe handles your payment details, and WhisperType never sees your card number. Stripe may collect billing, payment, tax, and fraud-prevention information under its own policies. WhisperType uses Stripe to manage purchases, subscription billing, and refunds.
5. How we use information
- To start and enforce the one-per-person free trial
- To activate and validate trials and subscriptions
- To provide customer support
- To process refunds, cancellations, and billing-related requests
- To diagnose technical problems if you contact support
6. Data retention
We keep only the minimum information required for trial enforcement, licensing, support, accounting, and legal compliance. Trial and purchase records may be retained as required by applicable law and by our payment provider.
7. International customers
WhisperType may sell to customers worldwide. Payment and tax processing may involve international providers. WhisperType is operated through Kenji Larrazabal (sole proprietor), Cusco, Peru.
8. Your rights
You can contact us to request access, correction, or deletion of personal data we control, subject to legal, fraud-prevention, accounting, and payment-provider obligations.
9. Contact
For privacy questions, support, or data requests, contact hello@whispertype.ai.
Company: Kenji Larrazabal (sole proprietor), Cusco, Peru.